High Performance Non-Oxide Ceramics I

101
Structure and Bonding
Managing Editor:
D.M.P. Mingos
Editorial Board:
AJ. Bard • I.G. Dance • P. Day • J.A. Ibers • T. Kunitake
TJ. Meyer -H.W. RoeskyJ.-P. Sauvage
A. Simon • F. Wudl
Springer
Berlin
Heidelberg
New York
Barcelona
Hong Kong
London
Milan
Paris
Tokyo
High Performance
Non-Oxide Ceramics I
Volume Editor: M. Jansen
With contributions by
F. Aldinger, S. Friihauf, U. Herzog, M. Jansen,
B. Jaschke, T. Jaschke, E. Miiller,
G. Roewer, H.J. Seifert, K. Trommer
Springer
The series Structure and Bonding publishes critical reviews on topics of research
concerned with chemical structure and bonding. The scope of the series spans the
entire Periodic Table. It focuses attention on new and developing areas of modern
structural and theoretical chemistry such as nanostructures, molecular electronics,
designed molecular solids, surfaces, metal clusters and supramolecular structures.
Physical and spectroscopic techniques used to determine, examine and model
structures fall within the purview of Structure and Bonding to the extent that the
focus is on the scientific results obtained and not on specialist information
concerning the techniques themselves. Issues associated with the development of
bonding models and generalizations that illuminate the reactivity pathways and
rates of chemical processes are also relevant.
As a rule, contributions are specially commissioned. The editors and publishers

will, however, always be pleased to receive suggestions and supplementary
information. Papers are accepted for Structure and Bonding in English.
In references Structure and Bonding is abbreviated Struct Bond
and is cited as a journal.
Springer WWW home page: http://www.springer.de

Visit the SB home page at http://link.springer.de/series/sb/ or
http://link.springer-ny.com/series/sb/
ISSN 0081-5993
ISBN 3-540-43131-4
Springer-Verlag Berlin Heidelberg New York
CIP Data applied for
This work is subject to copyright. All rights are reserved, whether the whole or
part of the material is concerned, specifically the rights of translation, reprinting,
reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in
any other way, and storage in data banks. Duplication of this publication or parts
thereof is permitted only under the provisions of the German Copyright Law of
September 9, 1965, in its current version, and permission for use must always be
obtained from Springer-Verlag. Violations are liable for prosecution under the
German Copyright Law.
Springer-Verlag Berlin Heidelberg New York a member of BertelsmannSpringer

Science + Business Media GmbH
http://www.springer.de
© Springer-Verlag Berlin Heidelberg 2002
Printed in Germany
The use of registered names, trademarks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are exempt from the
relevant protective laws and regulations and therefore free for general use.
Typesetting: Scientific Publishing Services (P) Ltd, Madras

Production editor: Christiane Messerschmidt, Rheinau
Cover: Medio Technologies AG, Berlin
Printed on acid-free paper SPIN: 107 19318 02/3020 - 5 4 3 2 1 0
Managing Editor Volume Editor
Professor D. Michael P. Mingos Professor Dr. Martin Jansen

Principal Max-Planck-Institut fur Festkorperforschung
St. Edmund Hall Heisenbergstrafie 1
Oxford 0X1 4AR, UK 70569 Stuttgart
E-mail: michael.mingos@seh.ox.ac.uk Germany
E-mail: martin.jansen@fkf.mpg.de
Editorial Board
Prof. Allen J. Bard Prof. Ian G. Dance

Department of Chemistry and Biochemistry Department of Inorganic and Nuclear Chemistry
University of Texas School of Chemistry
24th Street and Speedway University of New South Wales
Austin, Texas 78712, USA Sydney, NSW 2052, Australia
E-mail: ajbard@mail.utexas.edu E-mail: i.dance@unsw.edu.au
Prof. Peter Day, FRS Prof. James A. Ibers

Director and Fullerian Professor of Chemistry Department of Chemistry
The Royal Institution of Great Britain North Western University
21 Albemarle Street 2145 Sheridan Road
London WIX 4BS, UK Evanston, Illinois 60208-3113, USA
E-mail; pday@ri.ac.uk E-mail: ibers@chem.nwu.edu
Prof. Toyohi Kunitake Prof. Thomas J. Meyer
Faculty of Engineering: Associate Laboratory Director for Strategic and
Department of Organic Synthesis Supporting Research
Kyushu University Los Alamos National Laboratory
Hakozaki 6-10-1, Higashi-ku PO Box 1663
Fukuoka 812, Japan Mail Stop A 127
E-mail: kunitcm@box.nc.kyushu-u.ac.jp Los Alamos, NM 87545, USA
E-mail: tjmeyer@lanl.gov
Prof. Jean-Pierre Sauvage
Faculte de Chimie Laboratoires de Chimie Prof. Herbert W. Roesky
Organo-Minerale Institut fur Anorganische Chemie
Universite Louis Pasteur der Universitat Gottingen
4, rue Blaise Pascal Tammannstrafie 4
67070 Strasbourg Cedex, France D-37077 Gottingen, Germany
E-mail: sauvage@chimie.u-strasbg.fr E-mail: hroesky@gwdg.de
Prof. Fred Wudl Prof. Arndt Simon
Department of Chemistry Max-Planck-Institut fur
University of California Festkorperforschung
LosAngeles, CA 90024-1569, USA Heisenbergstrafie 1
E-mail: wudl@chem.ucla.edu 70569 Stuttgart, Germany
E-mail: simon@simpow.mpi-stuttgart.mpg.de
Editorial
We are pleased to announce that Professor Michael Mingos, University of

Oxford, UK. is to be our new Managing Editor for Structure and
Bonding:
Professor Mingos graduated at the University of Manchester and then
moved to University of Sussex for his doctoral work with Professor Joseph
Chatt. He extended his experience with postdoctoral positions at the
Northwestern University with Professor J.A. Ibers (1968-70) and at the
University of Sussex with Professor R. Mason (1970-71). Professor Mingos
held Lectureships at the Queen Mary College, London University; the
University of Oxford; and was a Fellow of Keble College, Oxford (1976-1992).
Subsequently he became Reader in Inorganic Chemistry at the University of
Oxford, was appointed to the Sir Edward Frankland BP Chair in Inorganic
Chemistry at Imperial College in London in 1992 and was Dean of the Royal
College of Science at the Imperial College (1996-99). Currently he is Principal
of St. Edmund Hall and Professor of Inorganic Chemistry at the University of
Oxford and also Visiting Professor at Imperial College. He was elected a
Fellow of the Royal Society in 1992 and holds Honorary Degrees at
Manchester and Sussex Universities. Among the many awards and prizes
he has received the following should be mentioned: Corday Morgan Medal,
the Chemistries of the Noble Metals Award, and the Tilden Medal of the Royal
Society of Chemistry, the Michael Collins Award for Innovation in Microwave
Chemistry, Wilhelm Manchott Prize, Technical University, Munich, The Julia
S. Edward C. Lee Memorial Lectureship and Prize, University of Chicago, and
an Alexander von Humboldt Stiftung Forschungspreis, which is held at
Heidelberg University.
He has served and is still serving on the Editorial Board of many prestigious
journals and has published more than 400 original papers and four
undergraduate textbooks.
Professor Mingos joined the Editorial Board of Structure and Bonding in
1983. Since then, he has given a tremendous input into this review series and a
great deal of effort into establishing and maintaining Structure and Bonding.
We wish him good luck and an abundance of new ideas for his increased
responsibilities.
VIII Editorial
We are always open to ideas and suggestions that would help Professor
Mingos make an even greater success of our series. Please do not hesitate to
send us your suggestions.
Dr. Marion Hertel

Senior Editor Chemistry
Springer- Verlag
Structure and Bonding
Also Available Electronically
For all customers with a standing order for Structure and Bonding we offer the
electronic form via LINK free of charge. Please contact your librarian who can
receive a password for free access to the full articles by registration at:
http://link.springer.de/orders/index.htm
If you do not have a standing order you can nevertheless browse through the
table of contents of the volumes and the abstracts of each article at:
http://link.springer.de/series/sb/
http://link.springer-ny.com/series/sb/
There you will also find information about the
- Editorial Board
- Aims and Scope
- Instructions for Authors
Preface
The nitrides and carbides of boron and silicon are proving to be an excellent
choice when selecting materials for the design of devices that are to be
employed under particularly demanding environmental and thermal condi-
tions. The high degree of cross-linking, due to the preferred coordination
numbers of the predominantly covalently bonded constituents equalling or
exceeding three, lends these non-oxidic ceramics a high kinetic stability, and is
regarded as the microscopic origin of their impressive thermal and mechanical
durability. Thus it does not come as a surprise that the chemistry, the physical
properties and the engineering of the corresponding binary, ternary, and even
quaternary compounds have been the subject of intensive and sustained efforts
in research and development.
In the five reviews presented in the volumes 101 and 102 of "Structure and
Bonding" an attempt has been made to cover both the essential and the most
recent advances achieved in this particular field of materials research. The
scope of the individual contributions is such as to address both graduate
students, specializing in ceramic materials, and all scientists in academia or
industry dealing with materials research and development. Each review
provides, in its introductory part, the chemical, physical and, to some extent,
historical background of the respective material, and then focuses on the most
relevant and the most recent achievements.
Since the degree of maturity reached by the materials considered is rather
varied, the focus of the respective reviews is also quite different. Thus for SiC
and Si3N4, the main emphasis is placed on processing and shaping, while for
BN its transformation to the cubic polymorph is a major concern, and, finally,
the report on the still rather young class of amorphous Si/B/N/C ceramics is
mainly devoted to aspects related to chemical syntheses and basic character-
izations. Inspite of the fact that many phenomena are dominated by kinetic
control, knowing the underlying thermodynamic equilibria is a crucial
prerequisite to any deeper understanding of the nitride and carbide based
materials discussed here. Therefore, a comprehensive and critically assessed
compilation of thermodynamic data and phase equilibria for the quaternary
system Si/B/N/C as well as its ternary and binary sub-systems has been
included as an introductory chapter preceding the reviews devoted to specific
materials and their properties.
Stuttgart, April 2002 Martin Jansen
Contents
Phase Equilibria in the Si-B-C-N System

H.J. Seifert, F. Aldinger 1
Silicon Carbide - A Survey of Synthetic Approaches, Properties

and Applications
G. Roewer, U. Herzog, K. Trommer, E. Miiller, S. Friihauf 59
Amorphous Multinary Ceramics in the Si-B-N-C System

M. Jansen, B. Jaschke, T. Jaschke 137
Author Index Volume 101 193
Subject Index 195

Contents of Vol. 102
High Performance Non-Oxide Ceramics

Volume Editor: M. Jansen
Boron Nitrides - Properties, Synthesis and Applications

R. Haubner, M. Wilhelm, R. Weissenbacher, B. Lux
Silicon Nitride Ceramics

G. Petzow, M. Herrmann
Hans Jiirgen Seifert1, Fritz Aldinger2
Max-Planck-Institut fur Metallforschung and Institut fur Nichtmetallische Anorganische
Materialien, Universitat Stuttgart, Pulvermetallurgisches Laboratorium, Heisenbergstrafie 5,
70569 Stuttgart, Germany
1
e-mail: seifert@mf.mpg.de
2
e-mail: aldinger@aldix.mpi-stuttgart.mpg.de
Literature information on phase equilibria and materials thermodynamics of the quaternary

Si-B-C-N system and its binary (Si-B, Si-C, Si-N, B-C, B-N, C-N) and ternary (Si-B-C, Si-B-N,
Si-C-N, B-C-N) subsystems are reviewed. Experimental phase diagrams and thermodynamic
data are compared with results from thermodynamic calculations according to the
CALPHAD method (CALculation of PHAse Diagrams). Calculated thermodynamic func-
tions, different types of phase diagrams and consistent Scheil reaction schemes for the
ternary and quaternary systems are presented. The influence of multicomponent phase
reactions on the development and application of refractory and hard materials of this system
are discussed and a model for the explanation of the thermal stability of Si-B-C-N precursor-
derived ceramics is introduced.
Keywords: Phase equilibria, Thermodynamics, Si-B-C-N system, Precursor-derived ceramics,

CALPHAD
1 Introduction 2
2 Stable Solid Phases in the Si-B-C-N System 5
3 Thermodynamic Data 5
4 Binary Subsystems 7
4.1 The Si-B System 7
4.2 The Si-C System 10
4.3 The Si-N System 15
4.4 The B-C System 19
4.5 The B-N System 24
4.6 The C-N System 28
5 Ternary Subsystems 28
5.1 The Si-B-C System 28
5.2 The Si-B-N System 32
5.3 The Si-C-N System 35
5.4 The B-C-N System 41
Structure and Bonding, Vol. 101

H.J. Seifert • F. Aldinger
6 The Quaternary Si-B-C-N System 45
7 Conclusion 50
8 References 51
1
Introduction
Silicon, boron, carbon and compounds of the Si-B-C-N system (e.g.'Si3N4, SiC,
B4+5C, BN) are the basis for the development of technically important
refractory ceramics and hard materials [1], Although numerous types of
processing have been developed in the past for their production, the powder
technology and sintering at temperatures up to 2500 K is the most common
for the fabrication of components. "Pressureless" sintering, hot pressing and
hot isostatic pressing, often in combination with the use of sintering additives
and reactive or non-reactive atmospheres, are applied for materials densifi-
cation and tailoring the properties for specific applications [2-4]. Because of
technical and economical limitations inherent to these conventional tech-
niques and, especially, due to the demand for new covalently bonded Si-B-C-N
materials there is an ongoing intensive development of alternative processing
technologies. The thermolysis of metallorganic precursors is not only valid
for the production of bulk ceramics but also of composites, fibers, coatings,
powders and infiltrations and the production of fiber reinforced ceramics
[5-13]. Other promising techniques are chemical vapor deposition (CVD)
[14, 15], arc discharging for the production of nanotubes and nanoparticles
[16] or self-propagating high temperature synthesis (SHS) for rapid materials
processing [17].
A large variety of complicated single- or multiphase microstructures results
from all these methods and requires a deep understanding of the underlying
multicomponent phase equilibria. Such information is also valuable to analyze
the individual or combined solubilities of B, C and N in liquid and solid silicon
and to find out which ceramic phases form when crossing the solvus of silicon.
The mentioned processing routes mostly do not result in equilibrium phase
assemblages and microstructures. Frequently, the materials are not fully
crystallized and consist of amorphous phases. Also, nanocrystalline and
turbostratic materials are produced. Some processing routes, such as the
thermolysis of precursors or the chemical vapor deposition, cause the
formation of metastable amorphous solid solutions (B-C-N, Si-C-N) typically
segregated into phase compositions related to equilibrium phases crystallizing
at elevated temperatures [18].
Such non-equilibrium stages as well as related high temperature reactions
during processing, machining and application can only be understood on the
basis of the underlying equilibria states. Also, the contacting of materials such
as SiC, BN, Si3N4 and/or B4C (e.g. in multilayer or fiber-reinforced materials)
necessarily requires a knowledge of the multicomponent phase equilibria

[19, 20].
The phase equilibria of the quaternary system Si-B-C-N are composed of the
pure components silicon, boron, carbon and nitrogen, the solid compounds
S13N4, SiC, BN and B4+(5C and the silicon borides SiBn, SiB6 and SiB3. The
corresponding phase compositions are indicated in the concentration tetra-
hedron as shown in Fig. 1. The significant homogeneity ranges of boron
carbide and the silicon boride SiBn are schematically indicated. The only
known stable ternary phases are the compounds /?-SiC2N4 and Si2CN4
produced from polymers [21]. These phases decompose at about 1173 K and
1273 K, respectively, and do not influence phase reactions at temperatures
higher than 1300 K. Reported compounds on the connecting line C-BN (e.g.
BC2N, BC3N [1]) are considered to be metastable phases.
Beside these unary, binary and ternary phases, /?-C3N4 has been theoret-
ically predicted by Liu and Cohen [22] as a solid phase but has not been found
experimentally yet. Numerous different metastable, amorphous phases of
binary, ternary or quaternary compositions have been obtained by gas phase
synthesis or polymer thermolysis. Bulk materials, fibers and coatings have
been produced by these methods [5-15]. To analyze the crystallization
behavior of such materials at high temperatures (T > 1800 K) a knowledge of
the phase equilibria with regard to the materials gross composition is required.
Theoretical work on metastable phases, e.g. Si3B3N7 [23], is also documented.
To evaluate the unary, binary, ternary and quaternary phase equilibria,
requires additionally taking into account the liquid phase and gaseous phases
Fig. 1. Scheme of the Si-B-C-N concentration tetrahedron with stable and metastable solid
phases indicated
(including numerous gaseous species). Although the metastable phases

mentioned above are not relevant to the stable phase equilibria, frequently
they play an important role in the understanding of the phase evolution during
the formation of the phase equilibria.
Ceramics of the Si-B^C-N system derive their outstanding properties from
the strong covalent bonds in the structure of the establishing phases. A special
feature of such inorganic covalent structures are low atomic mobilities,
providing an outstanding microstructure stability even at very high temper-
atures. These are advantageous properties for technical applications. On the
other hand, these properties cause specific difficulties in exploring the
materials chemistry and phase equilibria. Sluggish reactions even at temper-
atures higher than 1800 K complicate the preparation of well-defined samples
with equilibrium microstructures. Thermal analysis by DTA, DSC or TG is
difficult due to reactions with crucible materials and evaporation of boron and
silicon. The materials analysis by X-ray diffraction is complicated by intricate
patterns with overlapping and broadened reflections. Additionally, even minor
oxygen contaminations cause the formation of volatile species such as CO,
CO2, SiO and BO and therefore deviating sample compositions. In the past,
mainly problems such as these caused controversial discussions on phase
equilibria and stability of phases in the Si-B-C-N system and its subsystems.
Although there are significant improvements in analysis of light elements (e.g.,
EPMA), the characterization of materials in the Si-B-C-N system is still a
methodological challenge.
This paper gives an overview of experimental results and thermodynamic
calculations in the Si-B-C-N system and its binary and ternary subsystems.
In Sect. 2 some basic crystallographic data for the stable phases in the
Si-B-C-N system are given. Sect. 3 gives an overview on data sources for
thermodynamic calculations according to the CALPHAD approach (CALcula-
tion of PHAse Diagrams [24]). The following Sect. 4 describes the main
characteristics of the binary systems Si-B, Si-C, Si-N, B-C, B-N and C-N,
respectively, and Sect. 5 treats the ternary systems Si-B-C, Si-B-N, Si-C-N and
B-C-N. In Sect. 6 details on the quaternary system Si-B-C-N are given. Beside
the information on experimental phase diagram data, thermodynamic phase
diagram calculations (CALPHAD) are given together with a comprehensive
literature overview. Various diagrammatic representations of calculated phase
equilibria are provided and compared to experimental data. The calculated
temperatures and concentrations for invariant binary reactions and Scheil
reaction schemes of the ternary and quaternary systems are provided. The Scheil
reaction scheme gives a clear survey of phase equilibria in a ternary or
quaternary system [25]. It shows the invariant and univariant equilibria by
omitting the concentrations of the phases and keeping only the temperature axis.
Special attention is paid to the systems Si-C-N, B-C-N and Si-B-C-N,
respectively, since they are of importance for the properties of technologically
interesting precursor-derived Si-(B-)C-N ceramics. It will be illustrated how
phase diagram calculations are used to simulate and understand the materials
high temperature reactions and crystallization behavior of the intermediate
amorphous state.
Stable Solid Phases in the Si-B-C-N System
The compositions of solid phases appearing in the Si-B-C-N system are

indicated in the concentration tetrahedron as shown in Fig. 1. Some of them
appear in different modifications. The basic crystallographic data of stable
phases are given in Table 1.
The solid solution phases boron, boron carbide (B4+^C) and the silicon
borides (SiBn) SiB6, S1B3) show significant homogeneity ranges. Phase diagram
calculations including these phases require a thermodynamic modeling with
regard to their crystal structure.
Thermodynamic Data
The CALPHAD method of computer coupling of phase diagrams and

thermodynamics [24] was used to explore the phase equilibria in the Si-B-C-N
system. Analytical descriptions of the Gibbs free energies for all stable phases
and gaseous species of the system were established in the literature and by the
Table 1. Crystallographic data of solid phases of the system Si-B-C-N. The data are from
[26]. Data for SiBn from [27], c-Si3N4 from [28] and £-SiC2N4 and Si2CN4 from [21]
Phase Pearson Space Proto-type Lattice parameters, nm Comment
symbol group
a b c
Silicon cF8 Fdim C(dia.) 0.54309

/?-boron hRlll Rim jJ-B 1.09251 2.38143
Graphite hP4 P63/mmc C(gra.) 0.2464 0.6711
SiB2.89 hR!4 Rim B6P 0.6319 1.2713 74.3 at%B
SiB6 OP340 Pnnm SiB6 1.4397 1.8318 0.9911
SiBn hR12 Rim B 1.111 2.387 94.4 at%B
/?-SiC cF8 Fiim ZnS 0.43581 Type 3C
a-SiC hP12 Pe^mc SiC 0.30807 1.51174 Type 6H
(X-S13N4 hP28 P31c Si3N4 0.7818 0.5591
j8-Si3N4 hP14 P63/m Si3N4 0.7608 0.2911
c- Si 3 N 4 cF8 Fdim MgAl 2 O 4 0.776 15 GPa,
or Fdi 2000 K
B4C hR15 Rim B13C2 0.5617 1.2137
a-BN hP4 P63/mmc BN 0.2504 0.6661
/3-BN cF8 Flim C(dia.) 0.36155 6.6 GPa,
1723 K
jS-BN cF8 FAim ZnS 0.454 1.5 GPa,
1773 K
y-BN hP4 P63mc ZnS 0.2550 0.423 11.5-12 GPa,
293-873 K
/?-SiC2N4 Pn3m 0.61885 Stable below
1173 K
Si2CN4 Aba2 0.544 1.358 0.481 Stable below
1273 K
HJ. Seifert • F. Aldinger
present authors. The thermodynamic functions are derived in the course of the
so-called "thermodynamic optimization" [29, 30] where model parameters are
adjusted to various types of experimental data (phase diagrams and
thermodynamics) by using the least squares method of Gauss. The thermo-
dynamic data were stored in a computer database and the binary and ternary
system descriptions were combined for extrapolating calculations of the
quaternary system.
The thermodynamic descriptions are based on data for the pure elements Si,
B, C and N, respectively, as provided by Dinsdale [31] and stored in the SGTE
database [32]. The gas phase consists of numerous gas species. The most
important are listed in Table 2.
Their thermodynamic descriptions were also accessed from the SGTE
database [32]. Datasets for the binary systems Si-B, Si-C, B-C and B-N,
respectively, were derived by the authors and their coworkers and published
elsewhere [33-36]. Because of a lack of data and the probably small energetic
differences out of measurable quantities, a single analytical Gibbs-energy
description was used to describe a- and /?-SiC [35]. The thermodynamic data
for the binary system Si-N used in this work was taken from [37]. /?-Si3N4 is
the stable modification whereas a-Si3N4 is a metastable modification [38]. One
single description was used to describe the silicon nitride phases. The C-N
system is of relevance for the gaseous state.
The ternary system Si-B-C was optimized in the scope of our work on
the Si-B-C-N system [33, 34]. The other three ternary systems (Si-B-N, Si-C-N,
B-C-N) could be calculated comprehensively by thermodynamic extrapola-
tions since all solid phases have negligible ranges of solubility. Thermody-
namic data of the ternary phases /?-SiC2N4 and Si2CN4, are not known yet,
however they are reported not to be stable at the conditions treated here
(P = 1 bar, T > 1300 K) [21].
Table 2. Gas species in the Si-B-C-N system. Thermodynamic data were used from [32]
B C3N
BCi C4
BC 2 C4N
BN C4N2
B2 C5
B2C C5N
C C6N
CN C6N2
CN 2 -CNN C7N
CN 2 -NCN C9N
CSi N
CSi2 NSi
C2 NSi 2
C 2 N-CCN N2
C 2 N-CNC N3
C2N2 Si
C2Si Si 2
C3 Si 3
The descriptions of all ternary systems were combined in one dataset to

simulate the phase equilibria in the quaternary Si-B-C-N system. The results
of the thermodynamic calculations of individual systems are shown in the
corresponding sections. Thermodynamic models used are the Redlich-Kister
polynomial [39], extrapolations according to Muggianu et al. [40] and the
compound energy formalism [41] to describe the solid solution phases
jS-boron, SiBn, SiB6) SiB3 and B4+5C.
4
Binary Subsystems
4.1
The Si-B System
Since the early work of Moissan and Stock (1900) [42] on the synthesis of the
silicon borides SiB3 and SiB6, the Si-B system has been the subject of numerous
investigations. Experimental phase diagram data available on this system are
summarized in Table 3 [27, 43-61].
Comprehensive phase diagram investigations were published by Knarr
(1959) [43], Samsonov (1963) [46], Arabei (1979) [52], Lugscheider et al.
(1979) [53], Viala et al. (1980) [55] and Armas et al. (1981) [56]. Olesinski and
Abbaschian (1984) [62] presented a critical assessment and calculation of the
Si-B system.
Besides SiB3 and SiB6, a further boron-rich phase SiBn, isotypic to //-boron,
has been established [63] with 14 < n < 40 (93.3-97.6 at.% B at 1873 K [56,
69]). Armas et al. [56] as well as Male and Salanoubat [57] reported on a two
phase region between SiBn and /?-boron, whereas other authors treated SiBn as
a solid solution of silicon in /?-boron [55]. The homogeneity range of the
boron-rich solution phase SiBn is not well established but according to
Olesinksi and Abbaschian [62] "the existence of a stable, solid, rhombohedral
phase containing roughly 95 at.% B is rather indisputable". SiBn forms
peritectically at 2310 K.
Also, information on the extension of the homogeneity ranges of SiB3 and
SiB6 phases is controversial and still uncertain, mainly due to the difficulties
in phase separation and chemical analysis. Due to meager experimental
information, the SiB6 phase has been accepted as a stoichiometric phase
by Olesinksi and Abbaschian [62]. However, referring to the work of Vlasse
et al. (1986) [58] a small homogeneity (SiB5.7-SiB61, 85-85.9 at.% B) can be
assumed. SiB6 forms peritectically at 1658 K. For SiB3, Olesinski and
Abbaschian [62] accepted the results of Magnusson and Brosset (1962) [64]
and Ettmayer et al. (1970) [49], respectively, and gave the limits of the
homogeneity range as SiB2.89 and SiB 365 (74.3 to 78.5 at.% B). Similar values
for the homogeneity range were reported by Arabei (1979) [52]. Referring to
this phase various investigators reported a composition "SiB4" instead of SiB3.
This was explained by Olesinski and Abbaschian [62] by difficulties in phase
separation of SiB3 and SiB6 and therefore erroneous results. For further
discussion of the homogeneity ranges, see also Bolmgren (1994) [60]. SiB3 is
Table 3. Experimental phase diagram data of the system Si-B

Type of Type of Temp, Cone. Literature
experiment data range [K] range, x si
Metallography Microstructure T < 2373 0-1.0 Knarr (1959) [43]

Metallography Liquidus 1670-1800 0.8-1.0 Brosset and
Magnusson
(1960) [44]
Chemical Solvus 1433 0.99 Trumbore (1960)
analysis [45]
X-ray diffraction Solvus 1273-1573 0.97-0.99 Samsonov and
Sleptsov (1963)
[46]
Electrical Solvus 1173-1473 0.99-0.999 Hesse (1968) [47]
conductivity
Diffusion Solvus 1195-1420 0.996-0.997 Vick and Whittle
(1969) [48]
EPMA Si-B phases 1600-1873 0.8-0.9 Ettmayer et al.
(1970) [49]
Diffusion Solvus 1073-1286 0.998-0.999 Schwettmann

(1974) [50]
Metallography Solvus 1200-1573 0.992-0.998 Armigliato et al.
(1977), Masetti
et al. (1977) [51]
X-ray diffraction Si-B phases, 1873-2273 0-0.5 Arabei (1979) [52]
phase
compositions
X-ray diffraction, Solvus, liquidus 1473-1800 0.14-1.0 Lugscheider et al.
metallography, eutectic (1979) [53]
DTA equilibrium
Neutron Solvus 1073-1271 0.997-0.999 Ryssel et al.
activation (1980) [54]
X-ray diffraction Solvus 1873-2123 0.05-0.06 Viala et al. (1980)
[55]
Thermal analysis, Liquidus, solvus, 1700-2340 0-0.42 Armas et al.
X-ray SiBn phase (1981) [56]
diffraction
X-ray, Liquidus, SiBn 2340-2100 0-0.42 Male and
thermal analysis phase Salanoubat
(1981) [57]
X-ray Solidus, solvus 1670 0.14-0.15 Vlasse et al.
(1986) [58]
Electrical Solvus 1173-1473 0.995-0998 Borisenko and
conductivity Yudin (1987)
[59]
X-ray, EPMA Homogeneity 1123-1443 0.19-0.8 Bolmgren (1994)
range of SiB3 [60]
X-ray, EPMA Stability of SiB3 1498-1706 0.2-0.95 Aselage (1998)
[61]
accepted to be formed at about 1543 K by a peritectoid reaction [60]. Aselage

(1997) [61] reported SiB3 to be a metastable phase at all temperatures but
limited own experiments to temperatures T > 1498 K. Lim and Lukas [36]
calculated the temperature to 1543 K.
Besides the homogeneity ranges of the silicon borides mainly phase diagram
data concerning the liquidus and the silicon solvus were published. The
eutectic temperature between SiB6 and silicon was calculated to 1658 K [36]
(1657 K in the Scheil schemes) in good agreement with measurements with
experimental data [52, 56, 61].
Few experimental thermodynamic data are known for the Si-B system.
Relative enthalpies ("heat contents") [65], enthalpies of mixing of the liquid
phase [66, 67], enthalpies of formation data [68] and activity measurements
[43, 69-72] are reported. For an overview, see Table 4.
A first calculation of the equilibrium phase diagram of the Si-B system was
presented by Kaufman (1984) [73]. Olesinski and Abbaschian (1984) [62]
assessed and also calculated the phase diagram. To derive the analytical Gibbs
free energy descriptions for all phases, they took into account experimental
data on the phase diagram, thermodynamic and crystallographic data. The
diagram is in strong contradiction to the one presented by Kaufman [73]. The
system assessment of Olesinski and Abbaschian (1984) [62] was generally
accepted later by Dirkx and Spear (1987) [74]. However, these authors
established a new, thermodynamically optimized dataset for the system.
Additionally, in the frame of a thermodynamic assessment of the Si-B-C
system, Goujard et al. (1994) [75, 76] developed descriptions of the Gibbs
energies of formation for phases with the stoichiometric compositions SiB14,
Table 4. Experimental thermodynamic data of the system Si-B

Type of Temperature Compound, Literature
experiment range composition
[K] range
H(T)-H(298)
Drop calorimetry 80-1800 SiB4, SiB6 Bolgar et al. (1990) [65]
Enthalpy of mixing
Mixing calorimetry 1920 0-38 at.% B Esin (1978) [66]
Mixing calorimetry 1873 0-40 at.% B Biletskii et al. (1988) [67]
Enthalpy of formation
Vaporization experiments 298 SiB4, SiB6 Gordienko (1995) [68]
with mass spectrometry
Chemical potential
Knudsen cell 1473-2173 SiB6 Knarr (1959) [43]
Differential mass 1550-1840 SiBn-B Armas et al. (1981) [69]
spectrometry
Equilibrating 1723-1923 Dilute solution Noguchi et al. (1994) [70]
experiments of B in Si
Equilibrating 1723, 1773 Si-rich Tanahashi et al. (1998)
experiments [71]
Knudsen effusion 1522-1880 1.5-100 at.% B Zaitsev and
mass spectrometry Kodentsov (2000) [72]
10 H.J. Seifert • F. Aldinger
SiB6 and SiB3. They mainly used the activity measurements of Armas et al.
(1981) [69] to derive the data. A recent calculation of the Si-B system was
presented by Zaitsev and Kodentsov (2001) [72].
Lim and Lukas (1996) [36] took into account phase diagram and
thermodynamic data and presented a thermodynamically optimized dataset
for the Si-B system. Data for the homogeneity ranges of the intermediate
phases SiBn, SiB6 and SiB3, respectively, numerous data for the solubility of
boron in silicon (solvus, liquidus) (see Table 3) and thermodynamic data were
considered. Sublattice models related to the crystal structures were used to
describe the solid solution phases. Fig. 2a shows the optimized calculated
phase diagram in comparison with experimental data. The system shows
two perticectic reactions forming SiBn and SiB6, respectively, one eutectic
decomposition of the liquid phase and a peritectoid formation of SiB3 at
1543 K. From own experimental results Kasper [34] derived the peritectoid
temperature to 1471 K. This temperature is accepted in this work for the Scheil
reaction schemes as shown later for the Si and B containing ternary and
quaternary systems.
Phase diagram calculations (P = 1 bar) show that the gas phase appears at
temperatures as high as 3500 K. Fig. 2b shows the Si-rich side of the system.
The corresponding calculated invariant reactions are shown in Table 5.
4.2
The Si-C System
Silicon carbide (SiC) is the only solid compound known in this system. It
occurs with two modifications, cubic jS-SiC (referred to as "low temperature"
modification) and hexagonal a-SiC with numerous polytypes where 4H, 6H
and 15R are the most common ones. The a-//?-SiC transformation temperature
is not well known. Kistler-de Coppi (1985) [77] studied in detail the
transformation kinetics at temperatures between 1973 K and 2573 K. Accord-
ing to these results (metastable) /J-SiC transforms at temperatures above
2273 K to a-SiC. SiC melts incongruently forming a silicon-rich liquid phase
and graphite.
An overview of experimental phase diagram data is given in Table 6
[78-87]. A first assessment of the Si-C system was provided by Olesinski and
Abbaschian (1984) [88] mainly recommending the experimental phase
diagram work of DoUoff (1960) [80].
The peritectic melting temperature of SiC was set to 2818 K and the eutectic
temperature between (Si) and SiC to 1677 K. However, these data were
significantly corrected later by Kleykamp and Schumacher (1993) [85] due to
literature assessment and their own experimental results for temperatures
between 1700 and 3300 K. The peritectic temperature for SiC was measured to
occur at 3103 K ± 40 K. This temperature confirms the data given by Scace
and Slack (1959) [79] and Kieffer et al. (1969) [89]. The (degenerated) eutectic
point between (Si) and SiC was set to 1686 K + 1 K (melting point Si: 1687 K)
and 0.02 at.%. Additionally, the liquidus on the Si-rich side of the system was
measured. Data provided by Dolloff (1960) [80] scatter significantly from data
Phase Equilibria in the Si-B-C-N System 11
• [eOBro]
[81 Mai]
T [81 Arm]
2200 • [80V
• [79Ara]
21800-
1600-
0.4 06
Mole Fraction Si Si
Liquid
-
1600 - SiBg + (Si)

\
1543 \
\L (Si)
* • T
1400-
A
used:
[60Bro]
not used:
• [60Tru]
SiB 3 + (Si)
A
X\
• [69Vic] i [63Sam] A
• [74Sch] T [68Hes]
• [77Arm]
1200 - • [80Rys]
* [87Bor]
Male Fraction Si Si
Fig. 2a, b. Calculated Si-B phase diagram in comparison with experimental data [36].
a) Complete diagram, b) Si-rich side
Table 5. Invariant reactions in the system Si-B
Reaction Composition Composition Composition Temperature

at.% Si at.% Si at.% Si
1+(B) = SiBn (p) 7.4 2.1 3.3 2310

l+SiBn = SiB6 (p) 37.9 5.9 14.5 2123
1 = (Si)+SiB6 (e) 91.9 99.1 14.6 1658
(Si)+SiB6 = SiB3 (p) 99.5 14.5 26.2 1471
Table 6. Experimental phase diagram data of the system Si-C
Phase diagram
Type of experiment, Type of Temp. Cone. Literature

analysis data range [K] range, x c
Gravimetry Liquidus 1793-1998 0-0.01 Hall (1958) [78]

Gravimetry Liquidus 1833-3173 0-0.2 Scace and Slack
(1959) [79]
Thermal analysis, Liquidus, 1675-2950 0-1 Dolloff (1960) [80]
metallography, complete phase
X-ray diffraction diagram
Charged particle Liquidus 1685 0-0.01 Nozaki et al.
activation analysis (1970) [81]
Liquidus 1687 0-0.01 Voltmer and
Padovani
(1973) [82]
Combustion, Liquidus 1973-2423 0-0.01 Oden and McCune
IR-absorption (1987) [83]
Combustion, Liquidus 1708-1788 0-0.01 Suhara et al.
IR-absoption (1989) [84]
Thermal analysis, Peritectic 3113 0-0.2 Kleykamp and
chemical analysis equilibrium Schumacher
liquidus (1993) [85]
Combustion, Liquidus 1723-1873 0-0.01 Yanaba et al.
IR absorption, (1997) [86]
combustion,
coulorimetric
titration
Combustion, Liquidus 1720-1950 0-0.01 Ottem (1997) [87]
IR-absorption
given by Kleykamp and Schumacher (1993) [85] and many other published
data and seem to be influenced by systematic error. Due to the technical
importance, a lot of work was performed on the solubility of carbon in liquid
Si (Table 6, liquidus data). However, the scatter of data is significant. An
overview was given by Yanaba (1997) [86].
Literature sources of experimental thermodynamic data are listed in Table 7
[90-110]. A source of such information is also the review of Coltters (1985)
[111] and the compilation of Gurvich (1991) [112]. Low and high temperature
heat capacity measurements are available [90-94]. Heat contents [90, 95-97]
Table 7. Experimental thermodynamic data of the system Si-C
Type of Temperature Phase Literature

experiment range [K]
Heat capacities
Calorimetry 54-296 a-SiC Humphrey (1952) [90]
53-297 jS-SiC
54-295 a-SiC Kelley (1941) [91]
373-1073 Walker et al. (1962) [92]
5-300 jS-SiC Koshchenko et al.
(1979) [93]
5-2500 /J-SiC Koshchenko et al.
(1985) [94]
H(T)-H(298 K)
Drop calorimetry 298-1200 Not indicated Magnus (1923) [95]
298-1789 a-SiC Humphrey (1952) [90]
298-1693 jS-SiC
1365-2150 Not indicated Guseva et al. (1966) [96]
1100-2900 a-SiC Chekhovskoy (1971) [97]
1100-2900 j8-SiC
Combustion in 298 Not indicated Ruff and Grieger
Na2O2 (1933) [98]
Oxygen bomb a-, jS-SiC Humphrey et al. (1952)
calorimetry and JANAF [90, 140]
Oxygen bomb a-SiC Yudin and Markholiya
calorimetry (1969) [99]
Fluorine bomb a-, p-SiC Greenberg et al. (1970)
calorimetry [100]
Chemical potential
SiC solubility in 1563-1963 a-, P-SiC Chipman et al. (1954)
liquid Fe-Si alloy [101]
Knudsen cell, 2000-2300 a-SiC Drowart et al. (1960)
mass spectrometry [102]
Knudsen cell, 1808-1973 a-SiC Grieveson and Alcock
transp. technique (1960) [103]
Knudsen cell, 1808-1973 a-, j3-SiC Davis et al. (1961) [104]
weight loss
SiC solubility in Pb 1693 a-SiC Kirkwood and Chipman
(1961) [105]
SiC solubility in Ag 1693 /8-SiC D'Entremont and
Chipman (1963) [106]
SiC solubility in Fe 1823-1873 jg-SiC Rein and Chipman (1963)
[107]
SiC solubility in Au 1623-1898 a-, 0-SiC Sambasivan et al. (1993)
[108]
Knudsen cell, mass 1600-2000 a-SiC Rocabois (1995) [109]
spectrometry
Emf 1100-1300 a-, /J-SiC Kleykamp (1998) [110]
enthalpies of formation [90, 98-100] and chemical potential measurements

[101-110] are also reported. Comprehensive listings with enthalpies of
formation derived from 2nd and 3rd law data analyses can be found in
[110-112].
Numerous measurements of the chemical potential of silicon in the two
phase field carbon-SiC, i.e. the Gibbs energy of formation of SiC, are available.
A comprehensive discussion of these data can be found in Kleykamp (1998)
[110]. He stated that a-SiC is the more stable modification in the investigated
temperature range (1100-1300 K).
Calculations of the Si-C system were presented by Kaufman (1979) [113]
and Olesinski and Abbaschian (1984) [88]. Lim and Lukas (1996) [36]
published a thermodynamically optimized dataset which was refined by
Grobner (1996) [35]. This dataset is accepted by the present evaluators. The
accompanying calculated phase diagram is shown in Fig. 3. SiC decomposes to
graphite and carbon-rich liquid silicon at a temperature of 3095 K.
More recently published experimental data are also included in Fig. 3. The
data of Dollof [80] were not used for the optimization. However, Rocabois et al.
[109] concluded that their results from thermal analysis and silicon activity
measurements are more consistent to the data of Dollof [80] than to those
given by Kleykamp [85].
The solubility of carbon in solid Si is low and has not been taken into
account for the optimization. An overview of such measurements is given in
Liquid Liquid + Graphite
3095
Liquid + SiC
SiC + Graphite
used: T [58Hal]
• [59Sca]
A [87Ode]
Si + SiC 4 [93Kle]
• [97Yan]
not used: • [SODol]
0 0.4 1
Si Mole fraction C
c
Fig. 3. Calculated Si-C phase diagram in comparison with experimental data [35]
Olesinski (1984) [88] and Kleykamp (1993) [85], respectively. Vaporization

experiments show that gaseous species Si, Si2, Si3, Si2C and SiC2 are the most
relevant species in the gas phase [109]. Fig. 4 shows the chemical potential of
silicon in the two phase region SiC-C in comparison with experimental data.
The calculated invariant reactions are given in Table 8. Because of a lack of
data and due to the very small energetic differences between a- and /?-SiC,
respectively, a single analytical Gibbs-energy description was used to describe
both SiC-modifications.
4.3
The Si-N System
Experimental thermodynamic investigations of the Si-N system mainly

concern the decomposition behavior of silicon nitride, Si3N4, which is the
only stable compound in the system [114]. At normal pressure two
polymorphs of silicon nitride, a- and /?-Si3N4) respectively, were described.
These modifications show related structures with hexagonal symmetry. The
ot-Si3N4 to /?-Si3N4 phase transition was detected at temperatures higher than
1800 K. The transition is not reversible on heating or cooling and hence
/?-Si3N4 is referred to as the stable and a-Si3N4 as a metastable modification
[114]. Results from high temperature drop solution calorimety [38] support
this assumption. It is also suggested that the transformation requires a certain
amount of impurities (oxygen). A third polymorph, c-Si3N4 with a spinel
structure, is reported to be stable at pressures above 15 GPa and temperatures
exceeding 2000 K [28].
3000-
2500-
^ 1
^ 2000-
g 1500-
1000-
used: • [60Gri]
T [63d'E]
A |o:in--i]
500- • [9SKIt]
/ig; in kJ/mol
Fig. 4. Chemical potential of silicon in the two phase region SiC-C

Table 8. Invariant reactions in the system Si-C

at.% Si at.% Si at.% Si
G = l+C 58.4 78.7 0 3216

1+C = SiC (p) 82.7 0 50.0 3095
1 = Si, SiC (d) 100.0 100.0 50.0 1687
An assessment of the Si-N system was presented by Carlson (1990) [114].

Thermodynamic calculations were published by Kaufman (1979) [115],
Dorner et al. (1981) [116] and Weiss et al. (1981) [117] and later by Hillert
et al. (1992) [37]. Fig. 5 shows the calculated phase diagram at normal pressure
using the data of Hillert et al. (1992 [37]) for Si3N4.
Because of the only small energetic differences a single analytical Gibbs-
energy description was used to describe a- and /?-Si3N4, respectively and since
a-Si3N4 is metastable [38] it was not included in the phase diagram. Several
studies are concerned with the solubility of nitrogen in liquid and solid silicon
(Table 9, [71, 118-120]) which is small and was not taken into account for
thermodynamic calculations. Accordingly small is also the lowering of the Si
melting point.
The silicon-rich part of the phase diagram according to Yatsurugi et al.
(1973) [119] was accepted in the evaluation by Carlson (1990) [114]. Further
nitrogen solubility data were provided [71, 120]. Melting experiments were
carried out by Kostanovskii and Evseev (1994) [121] and Cerenius (1999)
[122].
Different types of experimental thermodynamic data are available for Si3N4
and are listed in Table 10. Data for heat capacities [94, 123-125], relative
3500
Gas
3000-
Si(l) + Gas
2500-
2114 K
I" 2000 Si 3 N 4 +Si(l)
5 1687 K
Gas + Si3N4
1500-
Si 3 N 4 +Si(s)
1000
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Mole Fraction Si
Fig. 5. The calculated Si-N phase diagram
Table 9. Experimental phase diagram data of the system Si-N
Type of Type of data Temp. Cone. Literature

experiment range [K] range, x N
Zone melting Solvus, liquidus 1687 0-0.0002 Kaiser and
experiments Thurmond
(1959) [118]
Zone melting Solvus, liquidus 1687 0-0.00012 Yatsurugi et al.
experiments (1973) [119]
Equilibrium N solubility 1723-1873 0-0.00004 Narushima et al.
experiments in liquid Si (1994) [120]
Laser heating Melting of Si3N4 1300-4000 Si3N4 Kostanovskii and
Evseev (1994)
[121]
Equilibrium N solubility 1723, 1773 0-0.00004 Tanahashi et al.
experiments in liquid Si (1998) [71]
Diamond anvil High pressure 2000-3700 0.571 Cerenius (1999)
cell laser heating melting of Si3N4 P = 37 GPa [122]
enthalpies [126] and the standard enthalpy of formation [38, 127, 128] are
documented. Several groups published data for the Gibbs free energy [125,
129-137]. However, the thermodynamic data are rather conflicting.
For a thermodynamic evaluation see also Durham et al. (1991) [138],
Gurvich (1991) [112] and Hillert et al. (1992) [37].
Numerous values for the standard enthalpy of formation of Si3N4 have been
published. A literature compilation [112] reports data for the standard molar
enthalpy of formation AfH1^ (298.15 K) for Si3N4 spanning a range from
-621 kj/mol to —842 kj/mol. Most of the data were derived from equilibrium
N2-pressure measurements and second and third law treatments. The
calorimetric data of Wood (1973) [127] were recommended [112]. However,
as outlined in detail by O'Hare et al. (1999) [128] these data are not reliable.
Rocabois et al. (1996) [125], based on mass spectrometry data, gave a range
of -(815 ± 12) < AfH°m (Si3N4, 298.15 K) < -(770 ± 12) kj/mol. Andrievskii
and Lyutikov (1996) [137] reported a value of -736.4 kj/mol derived from
Knudsen measurements. More recently, O'Hare et al. (1997, 1999) [128] and
Liang et al. (1999) [38] published new results on the enthalpy of formation for
a- and /?-Si3N4, respectively. Both groups state that no energetic distinction
between a- and /?-Si3N4 is possible. Liang et al. [38] found that a-Si3N4 is not
stabilized by impurities (e.g., oxygen) and is probably always metastable. The
calculated value accepted by the present authors according to Hillert et al.
(1992) [37] is AfH°m (Si3N4, 298.15 K) = -842 kj/mol which is in between the
values given by O'Hare et al. [128] and Liang et al. [38], respectively.
Data for molar Gibbs free energy of formation AfG°m (Si3N4, 298.15 K) for a-
and /?-Si3N4 are reported in the literature. The data derived from equilibrium
nitrogen measurements by Pehlke and Elliot (1959) [131] for a-Si3N4 were
selected by Hillert [37] to be reliable. Blegen [136] measured AfG°m for /?-Si3N4
and concluded that the stability difference between a- and /?-Si3N4 is small.
Table 10. Experimental thermodynamic data of the system Si-N
Type of Temperature Phase Literature

Heat capacities
Adiabatic 55-310 jS-Si3N4 with Guzman et al.
calorimetry a-Si 3 N 4 impurity (1976) [123]
Adiabatic 5-100 jS-Si3N4 with Koshchenko and
Calorimetry a-Si 3 N 4 impurity Grinberg (1982)
[94, 124]
DSC 220-620 Not specified Rocabois et al.
(1996) [125]
H(T)-H(298 K)
Drop calorimetry 530-2155 Not specified Neel et al. (1962)
[126]
Fluorine bomb 298.15 a-Si 3 N 4 Wood et al.
Fluorine bomb a-Si 3 N 4 , /5-Si3N4 O'Hare (1998,
calorimetry 1999) [128]
Drop solution a-Si 3 N 4) /5-Si3N4 Liang et al.
Vapor pressure and Gibbs free energy of formation
Heterogeneous 1700 a-Si 3 N 4 Matignon (1913)
equilibrium [129]
Heterogeneous 1606-1802 Not specified Hincke and Brantly
equilibrium, (1930) [130]
pressure measurements
Heterogeneous 1673-1973 GC-Si3N4 with Pehlke and Elliot
equilibrium, /?-Si3N4 impurities (1959) [131]
Heterogeneous 1618 a-Si 3 N 4 McClaine et al.
equilibrium, (1966) [132]
Knudsen and langmuir 1688-1773 Not specified Ryklis et al. (1969)
vaporization [133]
Heterogeneous 1473-1623 a-Si 3 N 4 Colquhoun et al.
equilibrium, pressure j3-Si3N4 (1973) [134]
measurements
Heterogeneous 1563 a-Si 3 N 4 Wild et al. (1970)
equilibrium £-Si 3 N 4 [135]
Heterogeneous 1683-1910 jS-Si3N4 Blegen (1975) [136]
equilibrium
Knudsen method 1993-2293 yg-Si3N4 Andrievskii and
Lyutikov (1995)
[137]
Knudsen cell, 1515-1690 Not specified Rocabois et al.
mass spectrometry (1996) [125]
The data of Blegen [136] were corrected later by Hendry (1977) [139]. Janaf
[140] gives a standard entropy value of 112.967 J/mol K whereas the low
temperature heat capacity measurements recommended by Koshchenko
[94, 124] give 64.2 J/mol K. This value was accepted by Hillert et al. [37].
At 2114 K and 1 bar total pressure Si3N4 decomposes into liquid silicon and
nitrogen gas according to the reaction:
Si3N4(s) = 2N2(g) (1)
According to the study of the vaporization behavior of Si3N4 [125] besides

N2 other important gaseous species are Si, Si2, Si3> SiN and Si2N. Heuer and
Lou (1990) [141] presented the volatility diagram and the Ellingham-type
representation for the Si-N system using thermodynamic data of JANAF [140].
The decomposition temperature of Si3N4 depends on the nitrogen pressure.
The corresponding potential phase diagram, calculated by using the data
of Hillert et al. [37], is shown in Fig. 6. Some decomposition data for Si3N4
are indicated. (p N2 = 10~4 bar, T = 1572 K; p N 2 = 1 bar, T = 2114 K;
p N 2 = 10 bar, T = 2307 K).
The melting temperature of a-Si3N4 up to a pressure of 37 GPa was
determined by diamond-anvil cell technique [122]. The melting temperature
varied from 2200 ± 75 K at 3.5 GPa to 3600 ± 200 K at 37 GPa. The authors
did not refer to a phase transition a-Si3N4/c-Si3N4.
4.4
The B-C System
In the B-C system one binary phase, boron carbide, is established. This phase
shows an extensive homogeneity range and to distinguish it from stoichio-
1-
^/2307K
0-
>^14K
Si 3 N 4
y
-2- -
-3-
y /
Si(l)
-4- -
y^1572K
/ Si(s)
-5-
1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400
Temperature [K]
Fig. 6. The Si-N potential diagram (log p N2 -T) [272]

metric B4C, notations such as (B4C), B13C2±X or B12(B, C)3 can be found. In the
present work it is indicated as B4+(5C.
Table 11 shows an overview of the experimental phase diagram data for the
B-C system [142-146].
The first published versions of the B-C phase diagram showed several boron
carbide phases on the boron-rich side [147, 148]. Additionally, Samsononv
et al. (1956) [147] reported a boron carbide phase to exist at about 50 at.%
carbon. Dollof (1960) [80] reported the existence of only one boron carbide
phase with an extended homogeneity range significantly narrowing at
temperatures lower than 2073 K. Elliot and van Thyne (1961) [142] compre-
hensively investigated the phase diagram from 1473 K to 2743 K and in the
concentration area from 0.75 to 40 at.%. One single boron carbide phase with
a significant homogeneity range from approximately 9 to 20 at.% carbon and
of congruent melting behavior (2723 K, 18.5 at.% C) was detected. No
temperature dependency of the homogeneity range was found. A degenerated
reaction between boron and boron carbide was documented at the compo-
sition and melting temperature of elemental boron (2348 K). An eutectic
reaction at 29 at.% carbon and 2648 K was mentioned. The data published by
Elliot and van Thyne [142] are generally accepted until today. With the same
analytical methods as Elliot and van Thyne [142] Kieffer et al. (1971) [144]
redetermined the phase diagram up to 60 at.% C. The results of Elliot and van
Thyne [142] were generally confirmed with deviations concerning the liquidus
in the boron-rich part and the eutectic reaction on the carbon-rich side. The
solubility of boron in graphite at temperatures from 2073 to 2773 K was
determined by Lowell (1967) [143]. Bouchacourt et al. [149] investigated the
boron-rich part of the phase diagram. They found a maximum carbon
solubility of 1 at.% in jS-rhombohedral boron and suggested a peritectic
reaction (1 + B4C = /?-B). The homogeneity range of boron carbide was
reported by Bouchacourt et al. (1981) [145] to extend from B 1 0 4 to B 40 C. An
overview on this work is given by Thevenot (1990) [150]. Beauvy (1983) [146]
found a maximum carbon solubility in boron carbide of 21.6 at.% at "low
temperatures" and 24.3 at.% near the eutectic temperature. This value differs
Table 11. Experimental phase diagram data of the system B-C
Type of Type of Temp. Cone. Literature

experiment data range [K] range, xc
Thermal analysis, Liquidus, solidus, 1473-2743 0-0.2 Elliot and Van
metallography, solvus Thyne (1961) [142]
X-ray
X-ray Solvus 1673-2773 0.97-1.0 Lowell (1967) [143]
Thermal analysis, Liquidus, solidus, 1473-2743 0-0.2 Kieffer et al. (1971)
metallography, solvus [144]
X-ray
X-ray Solvus 1800-2000 0.08-0.2 Bouchacourt and
Thevenot (1981) [145]
X-ray Solvus 1500-2723 0.18-0.24 Beauvy (1983) [146]
significantly from the results of Schwetz and Karduck (1991) [151] who used
EPMA and found a solubility limit of only 18.8 at.% (i.e., B4.3C) in the
temperature range from 2273 to 2623 K. The eutectic temperature of the
decomposition of the liquid phase to B4+(5C and graphite was detected to be at
2653 K in accordance with the value given by Elliot and van Thyne [142] and
Kieffer et al. [144]. Werheit et al. (1993, 1994) [152] found by lattice parameter
investigations a solubility of at least 1 at.% carbon in j8-boron at 1573 K. Some
selected experimental phase diagram data have been discussed by Okamoto
(1992) [153].
Different types of thermodynamic data were measured for the boron
carbide phase. An overview is given in Table 12 [91, 154-166].
In numerous cases the authors did not measure the exact composition of
the investigated materials but assumed a stoichiometric composition of "B4C".
Heat capacity data below room temperature (54.5-294.3 K) [91] and at high
temperatures (T > RT) [154-157] and relative enthalpy data of boron carbide
[158, 159] are available. The enthalpy of formation value (-11.6 ± 2) kj/mol of
atoms given by Smith et al. (1955) [160] determined by oxygen bomb
calorimetry was corrected by Domalski and Armstrong (1968) [161] to
-14.3 ± 3 KJ/mol of atoms. They used fluorine bomb calorimetry. Hong and
Kleppa (1978) [163] found by high-temperature solution calorimetry a similar
value of-14.0 kj/mol of atoms. Gal'chenko(1970) [162] found a similar value
but referred to a composition of B4.233C.
Froment et al. [164, 166] measured the carbon activities in the homogeneity
range of boron carbide by electromotive force method (emf, potentiometric
Table 12. Experimental thermodynamic data of the system B-C
Type of Temperature Literature

Heat capacities
DSC 54-294 Kelley (1941) [91]
DSC 323-973 Gilchrist and Preston (1979) [154]
DSC 300-700 Wood et al. (1985) [155]
Pulse calorimetry 300-1500 Matsui et al. (1991) [156]
Pulse calorimetry 30-300 Medwick et al. (1994) [157]
H(T)-H(298 K)
Drop calorimetry 430-1725 King (1949) [158]
Drop calorimetry 373-2600 Sheindlin et al. (1972) [159]
Combustion calorimetry 298 Smith et al. (1955) [160]
Calorimetry, F2 298 Domalski and Armstrong (1968)
[161]
Calorimetry 298 GaPchenko (1970)[162]
Solution calorimetry 1320 Hong and Kleppa (1978) [163]
Chemical potential
EMF 720 Froment et al. (1991) [164, 166]
Knudsen-cell Mass 2300 Froment et al. (1991) [165]
spectrometry
cells) at 620, 670 and 720 K, respectively. The data were discussed together
with the same type of data derived from Knudsen effusion mass spectrometry
(2300 K) applied by Froment et al. [165]. Based on the data provided by
Froment et al. [164-166] a thermodynamic description of boron carbide at
1300 K was derived by Goujard et al. (1994) [75, 76]. To describe a
homogeneity range "B 10 C-B 4 C" a sublattice model Ba(C, Va)c was used.
A thermodynamic optimization of the system was performed by Dorner
(1982) [167]. This dataset was later refined by Lim and Lukas (1996) [36]. Due
to additional crystallographic information concerning the extended homoge-
neity range of the boron carbide phase [152, 168] a further assessment was
necessary [33, 34, 169]. Data for the calculated invariant reactions are given in
Table 13. Boron carbide of composition 16.4 at.% C melts congruently at
2731 K.
The single phase region of B4+^C is arranged by atoms in icosahedral
geometry at the vertices of the rhombohedral unit cell and a three atom chain
on its main diagonal (parallel to the c-axis, hexagonal setting). The
distribution of carbon atoms on the boron carbide structure elements were
investigated quantitatively in dependence of the homogeneity range [152, 168].
The authors investigated the concentrations of the structural elements (B12,
B n C icosahedra and C-B-B and C-B-B chains) as well as the proportion of
chainless unit cells of boron carbide depending on the carbon-content. They
stated that nowhere in the homogeneity range the structure is uniform, rather
it consists of mixtures of B12 and B n C icosahedra and of mixtures of C-B-C
and C-B-B chains. Additionally, with decreasing carbon content a growing
proportion of unit cells without three atom chains develops. It was outlined
that by using IR-spectroscopy it is not possible to separate/divide between a
structural unit without any chain atoms and a unit with boron atoms at the
end positions and a vacant site at the middle (BVaB-type chain).
In orientation to the crystallographic results [152, 168], Kasper et al. (1996)
[33, 34] and Seifert et al. [169] used for the model description of
the homogeneity range of boron carbide the sublattice description
(B12) BnC)(CBC, CBB, BVaB). The sublattice model (B)93(B,C)12 was used to
describe the carbon solubility in /?-boron. The Redlich-Kister parameters for
the liquid phase and graphite (ss) and general formula descriptions for the
solution phases were accepted from [36]. The calculated optimized phase
Table 13. Invariant reactions in the system B-C

at.% C at.% C at.% C [K]
G=l 24.9 24.9 4258

g = 1+graphite 89.1 56.9 99.1 3918
1 = B4C 17.1 17.1 2734
1 = graphite+B4C (e) 29.1 97.9 18.9 2657
1+B4C = (B) (p) 0.35 8.8 1.5 2371
diagram is shown in Fig. 7a. The calculation on the boron-rich side is

compared with experimental data points in Fig. 7b.
Phase diagram data derived from contaminated samples (e.g. from reactions
with the crucible) or influenced by systematic errors in analyses (e.g. carbon
content measurements) were not taken into account for the optimization.
Good agreement with the most reliable experimental data is achieved by the
calculation. The calculated values give good agreement with standard entropy
data given by Kelley (1941) [91] and relative enthalpy data from King (1948)
[158] and Sheindlin (1972) [159].
a 4500
4000-
3500-
2 3000-
H
2500-
2000
1500
0.1 0.3 0.5 0.7
B Mole Fraction C
z
USED:
O [61 Ell] Eutect. R.
X [61E11] Solvus
V[61Ell]Liquidus
* [93Wer] Solvus
NOT USED:
X [71Kie] Inv. R.
*[71Kie] Liquidus
Y[61Ell]Liquidus
(BN-crucible)
G[61Ell]Solidus
N[81Boul] Solvus
«[83Bea] Solvus
A[91Sch] Solvus
1400
0 0.05 0.10 0.15 0.20 0.25 0.30 0.35
B Mole Fraction C
Fig. 7a, b. Calculated B-C phase diagram, [33]. a) Complete diagram, b) B-rich side.
Calculation in comparison with experimental data
24 HJ. Seifert • F. Aldinger
4.5
The B-N System
One solid compound, boron nitride (BN), exists in this system. No

homogeneity range was found. The compound occurs in four modifications.
Two graphite-like hexagonal (a-BN referred to as h-BN) and rhombohedral
(r-BN), and two dense, cubic (/?-BN referred to as c-BN) and wurtzitic (y-BN
referred to as w-BN) structures are documented (Table 1). Additionally,
turbostratic boron nitride (t-BN), with a highly distorted graphite-like
structure is known. No thermodynamic data are available on the CVD-derived
crystalline boron-rich boron nitrides of the compositions B2sN up to B53N and
B4N, respectively [170, 171].
A thermodynamic assessment and calculation of the B-N system at
normal pressure was presented by Wen (1994) [172]. Due to a lack of
experimental data a description for the liquid phase was chosen assuming
that the nitrogen solubility in liquid boron is comparable to the nitrogen
solubility in liquid aluminum. The homogeneity range of h-BN and /?-B is
experimentally not well established and stoichiometric descriptions were
selected. It was shown that the Neumann-Kopp rule is sufficient to
reproduce the experimental data for heat capacities and relative enthalpies.
Later, Kasper (1996) [33] accepted the data of Wen [172] for the gas phase
and the solid and liquid boron but took over the SGTE [32] description for
h-BN. This description fits slightly better to experimental thermodynamic
data of h-BN. The calculated phase diagram according to Kasper [33] is
shown in Fig. 8. h-BN is formed from the liquid and gas phase at 2767 K.
)3-B melts at 2348 K.
With respect to the technical importance of c-BN its stabilization by
pressurizing is of great interest. The first version of the BN pressure vs.
i i i i
4500
Gas
4000-
Gas + Liquid
3000-
2767
2500- Liquid + a-BN
2348 a-BN + Gas

2000- P-B + a-BN
1500
0.1 0.3 0.5 0.7 0.9
Mole Fraction N N
Fig. 8. Calculated B-N phase diagram at normal pressure [33]

temperature phase diagram was presented in 1963 by Bundy and Wentorf

[173]. This diagram was constructed from experimental data on the h-BN-cBN
transformation and melting of h-BN under high pressures. The diagram as
refined later by Corrigan and Bundy (1975) [174] was the generally accepted
version until the late 1980s. In the meantime the h-BN-cBN and vice versa
transformation was investigated by numerous groups. However, direct
methods and methods using catalysts interacting with BN, respectively,
resulted in a significant scatter of the results given for this transformation
([175], Table 14).
Solozhenko (1988) [176] concluded that kinetic factors influenced the
transformation significantly and that the true equilibrium line can only be
determined by a thermodynamic approach. From data on heat capacities
[177-189], relative enthalpies (heat contents) [190-199], enthalpies of
formation data [177, 201-207], equations of state and thermal expansion
data for all BN modifications, Solozhenko [176] derived a calculated new
phase diagram which significantly differed from the Corrigan and Bundy
(1975) [174] version. An overview of sources of thermodynamic data is
given in Table 15. A review of calorimetric studies was given by Gavrichev
et al. (1994) [212]. Vaporization studies of boron nitride were made by
[208-211].
The diagram was refined by Solozhenko in 1999 [175] using a new algorithm
(adapted pseudo-Debye model) for extrapolation of heat capacity data to
experimentally not known temperature ranges (T > 1650 K) and new exper-
imental data. For a comprehensive review on the B-N system at high
temperatures and high pressures we also refer to Solozhenko (1995) [213].
The pressure vs. temperature phase diagram according to Solozhenko [175]
together with other versions from the literature were shown by Rogl (2001)
[214] (Fig. 9).
A striking feature of the version according to Solozhenko [175, 176] is the
large region of stability of c-BN with normal pressure stability up to a
Table 14. Experimental phase diagram data of the system B-N

Type of Type of Pressure and Literature
experiment data temp, range
High pressure Triple point h-BN-c- 0-10 GPa; Bundy and Wentorf
experiments BN-liquid; h-BN- 298-4000 K (1963) [173]
c-BN transformation;
c-BN melting curve
High pressure Triple point h-BN-c-BN- 0-10 GPa; Corrigan and Wentorf
experiments liquid; h-BN-c-BN 298-4000 K (1975) [174]
transformation;
c-BN melting curve
Calculated from Triple point h-BN-c-BN- 0-10 GPa; Solozhenko (1988)
thermodynamic liquid; h-BN-c-BN 298-4000 K [176] Solozhenko
data transformation; (1999) [175]
c-BN melting; triple
point h-BN-liquid-vapor
Table IS. Experimental thermodynamic data of the system B-N
Type of Temperature Modification Literature

Low temperature heat capacities

Adiabatic calorimetry 20-300 h-BN Dvorkin et al.
(1954) [177]
h-BN Sirota et al.
(1975) [178]
5-300 h-BN Gorbunov et al.
(1988) [179]
15-305 High ordered Gavrichev et al.
h-BN (1993) [180]
5-300 c-BN Sirota et al.
(1975) [181]
4-302 Single crystal Solozhenko et al.
c-BN (1987) [182]
15-308 c-BN Gorbunov et al.
(1987) [183]
4-300 c-BN Solozhenko et al.
(1989) [184]
10-300 c-BN Atake et al.
(1990) [185]
5-320 w-BN Sirota et al.
(1976) [186]
6-307 w-BN Gorbunov et al.
(1988) [179]
15-305 r-BN Gavrichev et al.
(1992) [187]
15-305 t-BN Gavrichev et al.
(1993) [180]
High temperature heat: capacities
Estimation 1300-2200 h-BN Prophet and Stull
(1963) [188]
Scanning adiabatic 300-1100 Single crystals Lusternik and
calorimetry c-BN Solozhenko (1992)
M8Q1
[1O7J
Differential scanning 420-980 w-BN Solozhenko, unpubl.
calorimetry results
H(T)-H(298 K)
Drop calorimetry 298-1689 h-BN McDonalds and Stull
(1961) [190]
Drop calorimetry 300-1200 h-BN Agoshkov and
Bogdanova (1990)
[191]
Inverse drop 435-1652 h-BN Solozhenko (1993)
calorimetry [192]
Inverse drop 435-1524 r-BN Solozhenko (1993)
calorimetry [192]
Drop calorimetry 470-1238 c-BN Mezaki et al (1962)
[193]
298-1000 c-BN Kiseleva et al. (1973)
[194]
Table 15. (Contd)
Type of Temperature Modification Literature

experiment range

(1987) (s-crystal)
[195]
(1989) [196]
298-1200 c-BN Agoshkov and
Bogdanova (1990)
[191]
298-1500 c-BN Mayorova et al.
(1992) [197]
298-1700 c-BN Solozhenko (1993)
(s-crystal) [198]
300-1200 w-BN Agoshkov and
Bogdanova (1990)
[191]
Inverse drop 396-1287 w-BN Solozhenko (1993)
calorimetry [199]
Enthalpy of transformation
High-temperature 1800 c-BN-h-BN Solozhenko and
scanning heat flux Turkevich (1992)
calorimetry [200]
DSC, inverse drop 1500, w-BN-h-BN Solozhenko (1995)
calorimetry 1250-1400 [201]
Oxygen combustion h-BN Dvorkin et al. (1954)
calorimetry [177]
Reaction calorimetry h-BN Galchenko et al.
(1960) [202]
Fluorine bomb 298.15 h-BN Wise et al. (1966)
calorimetry [203]
Fluorine bomb h-BN Leonidov and
calorimetry Timofeev (1989)
[204]
Fluorine bomb c-BN Leonidov et al.
calorimetry • (1987) [205]
Evaluation from r-BN Solozhenko (1995)
w-BN-r-BN [201]
transformation
Fluorine combustion w-BN Leonidov et al.
calorimetry (1988) [206],
Leonidov and
O'Hare (1992)
[207]
temperature of about 1600 K. Previous work had stated an analogy of the BN

phase diagram to the carbon phase diagram, which was rejected in the work of
Solozhenko. Neither r-BN nor w-BN are stable phases in the investigated p-T
area. For further discussion of the phase diagram we refer to Rogl (2001) [214]
•—o Wentorf, 1959

Bundy, 1963
Corrigan, 1975
Rapoport, 1985
Solozhenko, 1988
Maki, 1991
Solozhenko, 1999
Kern, 1999, calcul
1000 2000 3000 4000 5000

Temperature [K]
Fig. 9. p-T phase diagram for the compound BN [214]
who points out that the in situ high pressure high temperature synchrotron
radiation data on the reversible h-BN-c-BN transformation provided recently
by Will et al. (2000) [215] "seem to some extent reconfirm the early
experimental diagram of Bundy and Wentorf (1963)" with a stability range of
h-BN extending at normal pressure as low as about 250 K.
4.6
The C-N System
Besides graphite no other solid phases are known in the C-N system at normal
pressure. The postulated compound C3N4 [22] was not found as a solid phase
until now. N2, CN and (CN)2 are the most important gas species.
Ternary Subsystems
5.1
The Si-B-C System
The Si-B-C system was mainly investigated with view to the understanding of
the sintering mechanisms of SiC with boron in combination with carbon and
the sintering of boron carbide with silicon [4]. The silicon solubility of about
2.5 at.% in B4C at 2323 K and the comparatively low temperatures of liquid
phase formation in the ternary system enhance the sintering of boron carbide.
The role of boron for favorable sintering of SiC is not completely solved.
However, a minor solution of boron in SiC seems to play an important role for
enhancing grain boundary diffusion [216].
The existence of pseudobinary sections was tracked by most of the groups
dealing with Si-B-C. Most of the authors assumed a composition B4+(5C with
<5 = 0. The phase equilibria of four different isopleths have been mainly
investigated experimentally in this system: B4C-SiC, B-SiC, B = const. = 80
at.% and B4C-Si. An overview of experimental investigations of the Si-B-C
phase diagram is shown in Table 16 [4, 217-224].
Calculated isothermal sections at 2500 K and 1400 K are shown in Figs. 10a
and b, respectively.
The calculated section B46C-SiC in comparison with experimental data is
shown in Fig. 11.
Secrist (1964) [218] reported a quasibinary eutectic reaction at 2573 K
whereas Shaffer (1969) [221] found a ternary eutectic reaction at 2518 K. Gugel
et al. and Kieffer et al. (1972) [223] reported a temperature of 2523 K. Instead
of eutectic reactions a transition reaction L + C = B4+^C + SiC at 2568 K is
calculated (Ul in Fig. 14). With respect to the scattering of the experimental
data this is a satisfying result. The solubilities of B4C in SiC and SiC in B4C
(equivalent to the solubility of silicon in B4C) seem to be lower than 3 mol.%
[218].
The experimental information for the solubility of boron in SiC is non-
uniform. Kasper (1996) [33, 34] accepted the solubility as given by Shaffer
(1969) [221] and described the solution phase with a substitutional model
where boron occupies the carbon positions.
The discussion in literature on the existence of the compound "SiB4" is
referred to in Sect. 4.1. Due to the documented isotypy of B4C and "SiB4" (i.e.-
the solid solution SiB2.89-3.6s) a complete solid solution at B = const. = 80 at.%
was assumed [219], However, due to the experimental work of Gugel et al. and
Kieffer et al. [223] and Telle [4] it is generally accepted today that no extended
solid solution exists. Telle [4] additionally rules out a complete solid solution
between SiB3 and B4C because of significant differences in the electron
configurations.
The section B4C-Si was investigated in [223]. Hot pressed samples treated at
2073 and 2273 K, respectively, were analyzed. All samples contained SiC and
B4C, respectively. The optimized ternary thermodynamic datasets of [36, 167]
were refined by Kasper in 1996 [33] and used for the calculation of the
B418C-Si section (Fig. 12). It shows that additionally to SiC and B4C a
silicon-rich liquid phase forms which was not found by [223].
The silicon solubility in B4+<5C was modeled by Kasper (1996) [33] by taking
into account Si2-units occupying C-sites in the linear C-B-C chain as described
by Werheit et al. (1994) [225]. The sublattice model for B4+gC (Sect. 4.4) was
extended to (B U C, B12)(CBC, CBB, BVaB, SiSi). The maximum solubility of Si
in boron carbide at 2323 K was found to be 2.5 at.% [4, 226]. This phase is in
equilibrium with boron saturated liquid silicon. The calculated liquidus
surface with isotherms and the traces of the univariant equilibria (1) 1, C,
B4+(5C (2) 1, C, SiC and (3) 1, B4+(5C, C are shown in Fig. 13.
Table 16. Experimental phase diagram data in the system Si-B-C o
Type of Type of Temp, range Section Literature

experiment data [K]
Density, electrical conductivity, Physical data, assumption 1973-2573, B4C-Si Portnoi et al. (1960) [217]
micro hardness of ternary phases hot pressing
X-ray diffraction, metallography, Phase equilibria, 2523, 2573, 2623 B4C-SiC Secrist (1964) [218]
chemical analysis eutectic reaction
X-ray diffraction, metallography Boron solubility 2273, 40 MPa B4C-SiC, B-SiC, Meerson (1966) [219]
in SiC hot pressing B = 80 at.% (const.)
X-ray diffraction, metallography Phase equilibria 1933 (B4C-SiC), B-SiC Dokukina et al. (1967) [220]
chemical analysis
Metallography Phase equilibria, eutectic 2473-2823 B4C-SiC, B-SiC Shaffer (1969) [221]
reaction, boron
solubility in SiC
X-ray diffraction Boron solubility 1933-1373 B-SiC Kalinina et al. (1971) [222]
in SiC
X-ray diffraction, thermal Complete ternary >1973 B4C-SiC B-SiC Gugel et al. (1972) [223],
analysis system B = 80 at.% B4C-Si Kieffer et al. (1972) [223]
X-ray diffraction, thermal Phase equilibria, silicon 2323 B-rich corner TeUe (1990) [4]
analysis, EELS solubility in B4C
X-ray diffraction Boron solubility Undetermined B-SiC Gierlotka et al.
in SiC (SHS) (1994) [224]
S.
T = 2500K
B4C + SiB
+ SiBn
B
B4C + SiB+ B
Fig. 10a, b. Calculated isothermal sections in the Si-B-C system, a) T = 2500 K, b) T = 1400 K
The transition reaction Ui (Fig. 14) appears at 2568 K. The deepest eutectic
temperature is calculated to be 1657 K. It is designated as degenerated reaction
D2 in the Scheil reaction scheme of Fig. 14. It is connected to the binary
eutectic reaction e3 (1 = Si + SiB6) but B4C does not participate in the ternary
reaction. The experimental value is 1658 K [4]. Three ternary transition
reactions and three degenerated reactions, respectively, occur. According to
the calculation six three-phase fields establish the equilibria at room
temperature.
A calculated isothermal section for the Si-B-C system at T = 1400 K was
presented by Goujard et al. (1994) [75, 76]. The data were derived from JANAF
[140] and assessments of the phases SiB14, SiB6, SiB3 and the solid solution
phase boron carbide. The isothermal section calculated using the data of
Kasper (1996) [33] (Fig. 10b) deviates significantly. Experimental investiga-
tions at these temperatures suffer from slow reaction kinetics and decisions
2800
2750-
0 20 40 60 80 100
B
4.6 C Mole Fraction SiC SiC
T [72Gug], Liquidus (Graphite furnace)
A [72Gug], Liquidus (Tungsten furnace)
• [64Sec], • [69Sha], not molten
® [64Sec], s [69Sha], partially molten
O [64Sec], • [69Sha], molten
Fig. 11. Calculated B46C-SiC section in the Si-B-C system. The horizontal line corresponds
to reaction U[ in Fig. 14 [33]
are difficult. However, the results of Kasper (1996) [33] are supported by data
given by Telle (1990) [4] concerning the sintering behavior of a sample of
composition 17.3Si72.2B7.5C (at.%) which is situated in the equilibrium
between B4+(5C and the ternary liquid. Additionally, at higher temperatures the
data of Kasper [4] are consistent with other experimental data. Goujard et al.
[75, 76] present one isothermal section at 1400 K and also find in some of the
CVD samples excess silicon which could be explained by phase relations given
in Fig. 10. For practical reasons it has to be emphasized that this isothermal
section at 1400 K is just one example and that other sections at any
temperature can be calculated using the thermodynamically optimized
dataset.
5.2
The Si-B-N System
Experimental work on the phase diagram of this system is meager. Some

information on selected phase equilibria can be derived indirectly from work
on the development of commercial materials in the Si-B-N system. Gugel et al.
(1972) [223] suggested a tentative phase equilibrium diagram. Equilibria
between silicon borides and silicon nitride were assumed. According to this
result, silicon does not show an equilibrium with BN. Kato et al. [227]
2600
2400-
2200-
2000-
1800-
1600-
1400
0 0 .1 0.2 0.3 0.4 0 .5 0.6 0.7 0.8 0.9 1.0
B
4.18 c X
Si Si
©Si-fSiC + Liquid
Fig. 12. Calculated B418C-Si section in the Si-B-C system. The horizontal line corresponds to
reaction U3 in Fig. 14 [33]
[K]
B 0.2 0.4 0.6 0.8

Mole Fraction C
Fig. 13. Calculated liquidus surface of the Si-B-C system [33]
investigated phase relations in the Si-Cr-N-B system at 1873 K and 1 atm N2.
For these conditions a tie line between Si3N4 and BN was established. No stable
ternary phase has been reported.
B-C C-Si B - C -Si B-Si

3095 I P.
l + C = SiC
2657 e,
B4C
2568 : = B4C +SiC U,
1
2371 C+B 4 C + SiC
I + B4C=B 2310 I
l + B = SiBn
2278 L + B = B4C + SiBn U;
I
HBJC+JSiBn
2123 I p.
2123 L+SiBn= SiB,, B4C D, I
I + SiB,, = SIB.
SiBn+SiB, + B(C
1687
I = Si, SiC
1669 L + SiC = B4C + Si U,
I
SiC + B4C + Si
1657 I e,
1657 L = Si + SiBe, B4C D,
I = Si + SiBs
1471 I p5
1471 |Si + SiB.= SiB3, B4CID,
Si + SiB. = SiB,
Si + SiB,+ B4C SiB. + SiB,+ B4C
Fig. 14. The Scheil reaction scheme of the Si-B-C system
Indirect phase equilibria information can be derived. Ceramic composites

Si3N4/BN were produced in the past by hot pressing [228] or precursor
thermolysis and subsequent hot pressing [229]. Such investigations show that
Si3N4 and BN are compatible up to the decomposition temperature of Si3N4
(nitrogen atmosphere). Noguchi et al. (1994) [70] measured activity coeffi-
cients of boron in a silicon melt and the interaction parameters of boron and
nitrogen in molten silicon. Solid BN and liquid silicon were equilibrated in a
nitrogen atmosphere from 1723 to 1923 K. Tanahashi et al. (1998) [71]
published potential phase diagrams (log pN2—log acB) of the Si-B-N system
valid for temperatures of 1723 and 1773 K, respectively. The combined
solubilities of boron and silicon in Si-B melts equilibrated with Si3N4 and
Si3N4/BN, respectively, were measured also at 1723 K and 1773 K and a
stability diagram was published.
Baldus et al. (1992) [230] produced a polymeric borosilicon imide amide by
ammonolysis of (trichlorosilyl)aminodichloroborane (TADB). This material
was heat-treated in an ammonia stream at 1273 K. An amorphous, single
phase boron silicon nitride with the formal composition Si3B3N7 was derived
when the powder was annealed in nitrogen at 1523 K. DTA-TG in helium
showed a decomposition reaction beginning at 1823 K and ending at 2053 K
resulting in Si, BN and N2. The decomposition is the result of the
decomposition of Si3N4. Materials pyrolyzed in nitrogen up to 2023 K were
shown to be amorphous to X-rays and electron beams.
The Si-B-N system was calculated in this work by extrapolation from the
binary subsystems. The isothermal section at 1273 K is shown in Fig. 15.
In agreement with experimental information at a pressure of 1 bar the
tieline Si3N4-BN exists up to the Si3N4 decomposition temperature of 2114 K.
The Scheil reaction scheme as derived from the calculation is given in Fig. 16.
Six invariant ternary reactions are calculated. All of them are of a degenerated
type because boron nitride, formed at 2767 K, does not participate in the
ternary reactions. At room temperature six three-phase equilibria are detected.
5.3
The Si-C-N System
Commercial materials in the Si-C-N system are usually developed by liquid

phase sintering (i.e. "pressureless" sintering, hot-pressing or hot-isostatic
pressing) of Si3N4/SiC powder mixtures with oxide additives [2, 3]. SiC-
reinforced Si3N4 composites are fabricated by addition of whiskers, fibers or
platelets of SiC to Si3N4. Crystallized materials with well-defined microstruc-
tures are produced. Amorphous or semicrystalline coatings and powders of
varying Si-C-N compositions are produced by chemical vapor deposition
(CVD) [231].
Additionally, the thermolysis of precursors has attracted wide attention
as an alternative approach to produce Si-C-N ceramics [6, 232]. At higher
temperatures this amorphous state is transformed into polycrystalline
ceramics consisting of thermodynamically stable phases, such as Si3N4, SiC
and graphite. Since the pyrolysis products can be used as both amorphous and
crystalline materials, the understanding of the high temperature behavior of
the ceramics and the phase reactions, the crystallization behavior and the
accompanying materials thermal degradation are of great importance with
O a-BN + siBj + SiB,

0 a-BN + SiB6 + SiB,
© a-BN + P-B + SiBn
0.4 0.6
Mole Fraction N
Fig. 15. Calculated isothermal section in the Si-B-N system (T = 1273 K) [33]
N-Si B-Si B -N -Si B-N

2767 1 p.
g + 1 = BN
2348 1 di
<ijnu | p2 2310 L + B = SiBn, BN Di
l + B - SiBn
B + SiBn + BN
2123 I P»
2123 L +SiBn = SiBs, BN D2
SiBn + SiBs+BN ,
2114 I P4
2114 G + L = SiaN4, BN D,
g + I = SisN4
I
G + BN + SisN4
1687 I d 2 1687 L = Si, SiaN4, BN D,

I - Si, SisN4
Si + SiaN4 + BN
1657 I ei L = Si + SiB., BN
1657 Ds
I = Si + SiBs
1471 P»
1471 Si + SiBe = SiBa, BN De
Si + SiB. = SiBs
I
c>i + SiBs+BN SiBs+S Bs + BN
Fig. 16. The Scheil reaction scheme in the Si-B-N system
respect to the physico-chemical properties of the polycrystalline ceramics and

to the maximum application temperature of the amorphous ceramics. This
information can be used for temperatures higher than about 1700 K where the
materials start to crystallize [233, 234]. At lower temperatures precursor-
derived samples maybe in a non-equilibrium state (amorphous, partially
crystalline) where the phase configuration cannot be reproduced by equilib-
rium calculations. However, NMR and wide angle X-ray scattering structures
[18] reveal that the as-thermolyzed amorphous states are segregated into
structural units typical for the proposed equilibrium phases.
Numerous thermodynamic calculations of phase equilibria in the Si-C-N
system have been published but only few experimental investigations are
documented. Calculated isothermal sections [117, 234-237], isopleths [117,
234], different types of potential phase diagrams [234, 235, 238-244] and phase
fraction diagrams [234, 239] were presented. Additional information is
provided by [245]. No or very low solid solubilities between SiC and Si3N4
could be detected by X-ray diffraction up to 2500 K [246, 247]. Also the
nitrogen solubility in SiC is low. For more experimental information see
[248, 249].
The Si-C-N system was calculated thermodynamically by extrapolations
from the datasets for the binary subsystems. The ternary phases Si2CN4 and
/?-SiC2N4 [21] are not stable at T > 1273 K and were not taken into account for
the calculations. Fig. 17 shows calculated isothermal, isobaric (P = 1 bar)

sections in the ternary system Si-C-N [234].
Silicon, graphite, SiC, Si3N4, the liquid and the gas phase (consisting mainly
of N2) establish the phase equilibria. Compositions of as-fhermolyzed
amorphous ceramics made from VT50- or NCP200- (i.e. PHMS-) precursors
and correlated reaction paths are indicated (arrows in Figs. 17b and c). The
production of these ceramics is described in detail in [250, 251]. The reaction
paths [252] indicate the change of the gross composition of the solid samples
due to the loss of nitrogen according to reactions (1) (Fig. 17b) and (2)
(Fig. 17c), respectively, as reported below.
Fig. 17a shows the phase equilibria for temperatures 1687 K <
T < 1757 K. At T < 1687 K silicon exists as solid phase. At 1757 K Si3N4
and graphite react according to the non-variant reaction
Si3N4(s) + 3C(s) =3SiC(s) + 2N2(g) (2)
In dependence on the thermodynamic data used, numerous different values

for the reaction temperature of reaction (2) were calculated: 1900 K [235],
1713 K [117], 1647 K [240], 1643 K [243] and 1757 K [244]. Experimental
Gas+Si,N.
SiC Si C SiC Si
Fig. 17a-d. Isothermal, isobaric sections in the Si-C-N system at 1 bar [234, 237]. The
compositions of the amorphous VT50- and NCP200- (PHMS-) derived ceramics (A, • ) and
reaction paths (arrows) are indicated, (a) 1687 K < T < 1757 K, (b) 1757 K < T < 2114 K,
(c) T = 2123 K, (d) T = 3000 K
"in situ measurements" of the temperature of reaction (2) are not available but
an extended heat treatment (in nitrogen atmosphere) of Si-C-N ceramics at
1773 K showed completely crystallized material consisting only of SiC and
carbon [253]. This result confirms that reaction (2) occurs close to the
calculated reaction temperature of 1757 K. Because of the sluggishness of the
reaction its temperature can hardly be detected exactly by thermal analysis
with defined heating rates [234].
The phase equilibria at temperatures between 1757 K and 2114 K are shown
in Fig. 17b. At 2114 K silicon nitride decomposes according to the reaction (1)
(see Sect. 4.3) into liquid silicon and nitrogen. Phase equilibria at a somewhat
higher temperature, T = 2123 K, are shown in Fig. 17c. Only two three-phase
fields, (a) gas + graphite + SiC and (b) gas + SiC + liquid, remain up to very
high temperatures where liquid silicon dissolves some carbon and the ternary
gas phase region is significantly extended (Fig. 17d, 3000 K). The gas phase
does not only consist of N2 but as well of other gaseous species such as Si, Si2,
SiC2, Si2C, CN and C2N2.
A Scheil reaction scheme was given by [117] and later modified with regard
to the reaction temperatures by [234]. The Scheil reaction scheme for the
Si-C-N system (valid for P = 1 bar) is shown in Fig. 18. Three degenerated
reactions and one transition reaction occur. The reactions (1) and (2)
C-Si Si-N C-Si-N N-C
3216 I e,
q = l+C
3095 P' - 3095 L + C-SiC, G

I + C - SiC
2114 P2 —| 2114 G + L = Si3N4 + SiC D*

g + I = SisN4
1757 G + SiC = C + SisN4 U,

I I
G + C + Si3N4 SiC + C + SisN4
1687 I di 1687 I d2 -=| 1687 L = Si, SiC, SisN. D,

1 = Si, SiC I - Si, SisN4
Si + SiC + SisN4
Fig. 18. Scheil reaction scheme for the Si-C-N system (/J-SiC/oc-SiC and a-Si3N4//J-Si3N4
transformations not indicated) [234]
correspond to Scheil reactions p 2 and Ui, respectively. The melting of Si at

1687 K causes the degenerated ternary reaction D3. At room temperature three
three-phase equilibria exist.
To simulate quantitative mass balances of the reaction paths and to simulate
thermogravimetric investigations phase fraction diagrams were calculated
[234, 237]. Fig. 19 shows this type of diagram calculated for a total pressure of
1 bar for the composition of the NCP200-derived ceramic with a ratio C:Si < 1,
where at the aforementioned temperature of 1757 K according to reaction
(2; Ui) 72 mass% Si3N4 and 8.8 mass% graphite form 29.4 mass% of SiC and
13.7 mass% of nitrogen.
Therefore, the predicted sample mass loss due to nitrogen evolution is
13.7%. The gas phase in this calculation still belongs to the system but appears
as "mass loss" in TG experiments. Due to the reaction all graphite is consumed
but excess Si3N4 (37.8 mass%) remains. At temperatures higher than 1841 °C
(2114 K) this residual Si3N4 decomposes according to reaction (1; p2) causing
a further nitrogen mass loss of 15.1%. The balanced reaction equation (2) is
only valid for a ratio C:Si3N4 defined by the intersection of the inner tielines
C-Si3N4 and N-SiC of the four-phase Scheil reaction Ux (79.6 mass% Si3N4,
20.4 mass% graphite). For precursor ceramics with lower carbon content
(e.g. NCP200) all carbon is consumed during reaction U^ and Si3N4 remains.
The calculated results were compared with experimental thermoanalytical
investigations [234, 237]. Fig. 20 shows the thermogravimetric curve for the
NCP200-derived ceramic.
Two steps of mass loss were detected as predicted by thermodynamic
calculation (Figs. 17b, c and 19). In the temperature range from 1500 °C
(1773 K) to 1690 °C (1963 K) 12% mass loss (calc. value: 13.7%) was detected,
and between 1840 °C (2113 K) and 1900 °C (2173 K) 15% mass loss (calc.
value: 15.1%) was determined. The two steps of mass loss agree quantitatively
well with the calculated ones. The temperature of reaction (1) is in agreement
80
1000 1500 2000 2500 3000 3500

Temperature [K]
Fig. 19. Calculated phase fraction diagrams in the Si-C-N system for PHMS-derived ceramics
(C:Si < 1) [234, 237]
1200 1300 1400 1500 1600 1700 1800 1900 2000

Temperature (°C)
Fig. 20. TG measurement (STA) for PHMS precursor-derived ceramic; N2 atmosphere,

BN-crucible, 10 K/min up to 1000 °C (1273 K), 5 K/min up to 2000 °C (2273 K) [234, 237]
with the calculated one and other experimental information [254]. The shift of
the reaction temperature (2) to higher values than calculated was explained by
kinetic hindrance [234]. Further sample analyses by X-ray diffraction (XRD),
scanning electron microscopy (SEM) and energy dispersive X-ray spectros-
copy (EDX) confirmed the calculated results [234, 237].
The influence of nitrogen gas pressure on Si-C-N phase reactions is shown
in Fig. 21 by the calculated potential phase diagram for Si-C-N materials
C:Si < 1 [244].
Along the upper line the three phases Si3N4, graphite and SiC are in
equilibrium and indicate reaction (2). Below the line the two phase field
SiC + Si3N4 (C:Si < 1) exists. Si3N4 decomposes at low partial pressures of N2
according to reaction (1) (Sect. 4.3) into nitrogen and liquid or solid Si which
is in equilibrium with SiC.
The potential phase diagram in Fig. 21 gives advice for the sintering of
Si3N4-SiC samples [239, 244]. If the N2-partial pressure is above the line
Si3N4 + C + SiC the sample is stable at the sintering temperature. For
sintering of SiC + Si3N4 there are temperature-dependent upper and lower
limits for the N2 pressure for the stability range. Nickel et al. [239] determined
the SiC whisker degradation and Si3N4-matrix decomposition by this diagram.
During cooling from the sintering temperature for such samples the N2
pressure has to be lowered simultaneously to keep the sample in the
SiC + Si3N4 phase field (Fig. 21). For more details see references [239, 241,
244]. Note the change of the reaction temperatures with the changing the
partial pressure of nitrogen. The reaction temperatures given in the Scheil
reaction scheme (Fig. 18) are valid for a total pressure of 1 bar. To evaluate the
stability of Si3N4 and SiC, respectively, Wada et al. (1988) [240] and Li et al.
(1995) [243] calculated various potential phase diagrams for different carbon
activity levels (a c = 1 and ac = 10~4) and nitrogen pressures.
1-
|C:SK1|
1 bar N 2
C+Si3N4
-1-
5
/ SiC+Si3N4
-2-
£ /
3 -3- / -Si(l)
/
1 Si(s)=Si(l)
-4-
/
J/
ySiC+sis) 1757 K 2114 K
-5-
1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200
Temperature [K]
Fig. 21. Potential phase diagram for the Si-C-N system; C:Si < 1 [234, 244]
5.4
The B-C-N System
No systematic experimental investigations of phase equilibria in this system

are reported. However, numerous approaches to produce different types of
commercial materials (bulk, coatings, nanotubes) are documented from which
some phase equilibrium information can be derived. Mainly materials of the
sections B4+<5C-BN and a-BN-graphite were investigated. Besides the hot
pressing method, CVD, precursor pyrolysis and arc discharging were used for
materials synthesis. The resulting boron carbonitrides BxCyNz can be treated
as metastable solid solutions of carbon in boron nitride with a structure
similar to graphite [255]. Due to the similarity of the crystal structures of a-BN
and graphite the possibility of solid solutions between both phases was
assumed. However, numerous experimental investigations showed all solid
solutions to be metastable.
Table 17 gives an overview of selected papers describing the experimental
work [14, 256-264].
The solubility of nitrogen in B4C was measured by Oscroft and Thompson
(1991) [256]. Fast-flowing ammonia (12 h) was used in the temperature range
from 1073 to 1373 K. X-ray diffraction, chemical analysis (SEM/EDX),
combustion analysis were used and a solubility of 3.8 mass% was found.
However, it can be assumed that the use of ammonia with high nitrogen
activities may have resulted in a metastable solid solution BxCyNz.
Ruh and Kearns (1992) [257] produced B4C-BN samples by vacuum hot
pressing at 2173, 2373 and 2523 K and found by lattice parameter analysis at
Table 17. Experimental phase diagram data in the system B-C-N
Type of Type of Temp. Section Literature

experiment/ data range [K]
analysis
X-ray diffraction Solubility of a-BN 1073, 1373 (B*C) Oscroft and

chemical in boroncarbide (NH3) sintered Thompson
analysis inNH 3 (1991) [256]
Phase equilibria, 2173, 2373, BN-(B4C) Ruh et al.
solid solubilities 2523, HP (1992) [257]
Solid solubilities 2173 (CVD), BN-C Badzian
2473, 2773 (1972) [258]
(3 GPa)
Phase equilibria 3600, BN-C Badzian (1981)
at high pressure 14 GPa [259]
1973 (CVD) BN-C Chen and
Diefendorf
(1980) [260]
1173 (CVD) BN-C Kaner et al.
(1987) [261]
2473 (precursor BN-C Bill et al.
thermolysis) (1992) [262]
Phase equilibria 1273 (CVD) BN-C Saugnac et al.
(metastable) (1992) [263]
1123-1448 BN-C-BC3 Derre et al.
(CVD) (1994) [264]
1123-1448 BN-C-BC3 Filipozzi et al.
(CVD); 2773 (1985) [14]
(heat
treatment)
T = 2523 K a "slight solubility of B4C in h-BN but no solubility of h-BN in

B4C". On the other hand Sirota et al. (1977) [265] found 7 mol-% boron
carbide solubility in cubic boron nitride. The samples were prepared by hot
pressing BN-B4C powder mixtures at 50 kbar and 2173-3073 K.
After the pioneering work of Badzian et al. (1972) [258] on the production
of graphite-boron nitride solid solutions by chemical vapor deposition
numerous similar investigations were published. Thin films of various
composition such as BC2N, B2CN2 and B0.35C0.3N0.35 were produced by CVD
experiments. For reviews see [14, 263, 264]. The films from CVD experiments
(1273-1423 K, boron trichloride, ammonia, and acetylene diluted in hydro-
gen) produced by Filipozzi et al. [14] show compositions inside the triangle
bounded by pure carbon, BN and "BC3". Typically hexagonal turbostratic
structures of varying quality of structural organization were found. X-ray
diffraction results show that the higher the carbon content the better is the
three-dimensional ordering of the graphitic structure. Subsequent heat
treatments (graphite resistor furnace, Ar-atmosphere, 1 bar) resulted in
"graphitization" of the layers. Up to 2000 °C only metastable single phase solid
solutions were found which was also confirmed by Raman microprobe studies.
Between 1750 and below 2500 °C some hints for boron carbide were found
and above 2500 °C only two phases were found: boron-saturated graphite and
"eventually BN" existed.
Amineboranes were thermally decomposed at 1273 K in Ar to boron
carbonitrides [262], These materials were reported to show a turbostratic
structure where atoms in a carbon-like structure are partially substituted
by boron and nitrogen, respectively. At temperatures higher than 2073 K
(Ar-atmosphere) the single phase material of compositions BC4N and BC2N,
respectively, started to crystallize and a-BN, B4C and elemental C (graphite)
were observed.
Sauter (2000) [266] investigated the structure of amorphous boron
carbonitride of composition B31C37N32 by X-ray and neutron diffraction as
well as by NMR spectroscopy. It was shown that annealed precursor-derived
B-C-N ceramics (1200 K < T < 1600 K) contain predominantly tricoordinated
boron, carbon and nitrogen atoms arranged in hexagonal rings, or fragments
of them, as structural units. No phase separation occurred.
Suenaga et al. (1999) [16] produced nanoparticles and nanotubes by arc
discharging and found by means of electron energy loss spectroscopy (EELS)
that such materials consist of immiscible BN and C-layers. The growth
mechanism was explained by using the calculated phase diagram sections in
the B-C-N system according to [33, 244].
Solozhenko et al. (1997) [267] carried out high pressure investigations of
metastable graphite-like BC4N. In situ powder diffraction of synchrotron
radiation was used for sample characterization up to 7 GPa and 2100 K. At
ambient pressure it was found that BC4N starts to decompose at 2050 K and
h-BN and disordered graphite are formed. With increasing pressure the
decomposition temperature decreases. Various other published works on
high-pressure, high-temperature investigations to produce dense ternary
phases in the boron-carbon-nitrogen system are cited in [267]. The synthesis
and structure of three-dimensionally ordered graphite-like g-BC2N under
high-temperature, high-pressure conditions (1200-1500 °C, 3-5 GPa) were
described by Nicolich et al. [268]. From graphite-like BC2N Solozhenko et al.
(2001) [269 and literature cited therein] synthesized superhard cubic BC2N.
The ternary system was calculated by extrapolation from the binary
subsystems (Kasper, 1996) [33]. The calculations cover phase equilibria at one
bar and do not assume any solubilities as no experimental evidence for stable
solid solutions between B4+(5C and BN or a-BN and graphite exist. The section
between graphite and boron nitride including the invariant reactions Uj, Dj
and U2 (Fig. 23) is shown in Fig. 22. A calculated potential phase diagram
(logpN2-T) can be found in [244]. The complete Scheil reaction scheme
(P = 1 bar) is shown in Fig. 23.
One maximum where boron carbide is formed from gas and liquid phases
was found by calculation. Two transition reactions and two degenerated
reactions were found. The calculated temperature of the reaction
B4C + 2 N 2 = 4 BN + C (U2 in Fig. 23) of 2597 K is experimentally not well
established. In situ investigations of the reaction of BN and graphite are not
reported and therefore only reaction temperatures indirectly derived from
X-ray analysis of quenched materials are documented. It has to be assumed
2900-
Gas + Liquid
Gas + Liquid + Graphite
2722
2700- 2657
2597 Gas + Graphite + B4tsC
3 2500-
CL Graphite + a-BN
P 2300 O Gas + Liquid + B44SC

© Gas + Liquid + a-BN
© Gas + Graphite
© Gas + Graphite + a-BN
2100- © Gas + B4rtC
c 0.2 0.4 0.6 0.8 BN
Mole Fraction BN
Fig. 22. Calculated temperature-concentration section C-BN in the B-C-N system [33]
B-C C-N B-C-N B-N

3918 I ei
g=i+c Z/b/ | P'
2733 I Max 1 + g = BN
G + L = BiC 1
r|2722 G + L = BN + B.C u,|
1
2657 I ez - 2657 G, L = B<C + C D,
I = C + B4C
2597 G + B4C = C + BN U2
G + C + BN B4C H C + B N
1
2371 Pa 2371 L+BiC = B, BN Dz
I + B4C = B
I 2348 d.
B4C + B + BN 1 = B, BN
Fig. 23. The Scheil reaction scheme of the B-C-N system
that the temperature is lower than the one given by Fillipozzi [14]
(T > 2723 K) and higher than the one given by Badzian [258] (2473 K) or
Bill [262] (2473 K). At room temperature three three-phase equilibria are
found.
6
The Quaternary Si-B-C-N System
Phase equilibria in this quaternary system are of main interest for the
understanding of the high-temperature behavior of precursor-derived
Si-B-C-N ceramics. An overview on their production and high-temperature
stability is given, e.g. in [6, 9, 13, 270]. The correlation of microstructure,
phase reactions and high-temperature stability for selected quaternary
precursor ceramics was discussed by Seifert et al. [236, 244, 271-273].
Interfacial reactions between silicon nitride and boron carbide processed
under nitrogen atmosphere were thermochemically calculated by Wiedemeier
and Singh (1989) [19]. The influence of nitrogen pressure and temperature on
the reaction Si3N4 + 3 B4C + 4 N 2 (g) = 3 SiC + 12 BN was investigated. It
was concluded that the SiC/BN system is stable up to temperatures of about
1800 K whereas the Si3N4/B4C system is unstable. The compatibility of SiC
with BN was confirmed by investigation of the interfacial reactions of BN-
coated SiC fibers at 1723 K [274]. However, Lee and Jacobson (1995) [20]
found experimental indications for a reaction zone at BN-SiC interfaces.
Si-B-C-N phase equilibria were calculated by extrapolating from the related
unary, binary and ternary systems into quaternary compositions [236, 244,
271-273]. This is valid since no solid solubilities and no quaternary phases
exist in the quaternary system. Calculated phase equilibria in the Si-B-C-N
system were shown exemplarily by isothermal sections at 1673 K and 2273 K,
respectively (Figs. 24a and b).
The concentration tetrahedra are divided into distorted tetrahedra indicat-
ing the 4-phase equilibria. Concentration sections through the 1673 K and
2273 K tetrahedra at a constant level of 10 at.% boron are shown in Figs. 25a
and b, respectively.
In these sections 4-phase equilibria appear as triangles or squares
depending on the way of cutting the tetrahedra. The 10 at.% section was
selected because compositions of numerous precursor-derived Si-B-C-N
ceramics are located within this section. Some compositions are indicated.
They are located in the phase field Si3N4 + SiC + C + BN. This phase
equilibrium remains stable as low as room temperature. However, at high
temperatures (e.g. 2273 K) the compositions are located in the phase field
Gas + SiC + C + BN and some decomposition of the material is expected.
Further calculated sections with numerous different precursor compositions
are discussed in [270].
The Scheil reaction scheme for the quaternary system valid for a total
pressure of 1 bar is shown in Fig. 26. It is consistent with the ternary Scheil-
reaction schemes as shown in Figs. 14, 16, 18 and 23. Twelve quaternary
invariant reactions occur: Besides the "maximum" (M) all of the reactions are
degenerated reactions where the fifth phase does not participate in the
reaction. At room temperature nine four-phase equilibria occur.
The phase fraction diagram for the high temperature stable PHBS(p)-
derived ceramic (composition indicated in Fig. 25) was calculated and is
shown in Fig. 27.
1673 K
SiB™
SiB,,
SiC
Si3N4 + SiC + Si + BN B4C + L + BN + SiB6

C + SiC + Si3N4 + BN B4C + BN + SiB6 + SiB,,
C + Si3N4 + Gas + BN B4C + B + SiBn + BN
B4C + SiC 4-BN + C L + S i + BN + SJC
B4C + L + E SiC
2273 K
SiBn
B4+BC
C, BN, SiC, B4C (L, BN, SiC, B4C)

B, BN, SiBn, B4C (G, L, SiC, BN)
C, G, BN, SiC (L, B4C, SiBn, BN)
Fig. 24a, b. Isothermal sections in the Si-B-C-N system [273]. (a) T = 1673 K, (b)
T = 2273 K (three of totally six 4-phase equilibria are shown)
As in the case for precursor-derived Si-C-N ceramics (see Sect. 5.3, Fig. 19)
reaction (2) (see Sect. 5.3 and reaction D6 in Fig. 26) appears at 1757 K. Si3N4
reacts completely with graphite to form SiC and gas phase. Excess graphite
remains. The BN phase does not take part in reaction (2; D6) and its amount
remains stable up to a temperature of 2586 K. From this result a thermal
decomposition of the PHBS(p)-derived material with significant mass losses is
expected. However, thermogravimetric analysis shows remarkable thermal
• PMBS(m)
• PHBS(p)
ffl PHBS(m)
• PMBS(p)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Mole Fraction Si Mole Fraction SI
Fig. 25a, b. Concentration sections in the Si-B-C-N system at constant boron content
(10 at.%) and temperatures. Composition of some high-temperature stable precursor-
derived ceramics are indicated [272, 273]. (a) T = 1673 K, (b) T = 2273 K
stability of this material and of various other precursor-derived boron-

containing ceramics up to 2300 K [270].
To understand these conflicting results the Si-B-C-N materials metastable
microstructure has to be taken into account. High resolution transmission
electron microscopy (HRTEM) reveals that the high-temperature stable
materials consist of nanocrystalline Si3N4 grains even in materials heat treated
at temperatures higher than the decomposition temperature of Si3N4 (2114 K).
These grains and such of SiC are embedded in a matrix of turbostratic B-N-C
layers [275]. Such turbostratic layers are well known from pure B-C-N
materials (see Sect. 5.4). Analyses of the B-N-C layers in PHBS(p)-derived
ceramics with electron spectroscopic imaging (ESI) shows dissolved carbon
with a B:N:C ratio of 1:1:3.3 [270]. The calculated B:N:C ratio can be derived
from the (BN + C) fraction in Fig. 27. The boron and nitrogen amount,
respectively, is 9 at.% (note the mole of atoms units), the graphite fraction is
28 at.%. This result is in very good accordance with the experimentally derived
ratio.
To explain the high-temperature stability with regard to the microstructure
of high-temperature stable precursor-derived Si-B-C-N ceramics an "encap-
sulation effect" of Si3N4 and SiC grains by the B-N-C matrix has been assumed
[244] which results in a diffusion barrier for nitrogen, a correlated internal
pressure increase and the high-temperature materials stabilization. To show
the influence of increased pressure on the decomposition and reaction
temperatures of Si3N4 calculated potential phase diagrams were used (Figs. 6
and 28) [244, 272, 273]. Fig. 6 shows that the Si3N4-decomposition temper-
ature according to reaction (1) increases from 2114 K at a total pressure of
1 bar to a temperature of 2307 K for a total pressure of 10 bar. Potential phase
diagrams were calculated for fixed Si:B:C ratios and variations of the nitrogen
potential, i.e., the nitrogen pressure (Fig. 28). To calculate this diagram the
B-C-N
2722 | ui
| G + L = BN + B*C
2657 I AT~
G + L = C + B.C
2597 I U2
G + B.C = C * BN
G, L, C, G, L, I 2590 I
BN SiC, BN||G = L+SiC + BN
125861 G + L = BN *
G,L,
L, BN, G, BN,
SiC, BN
I SiC, C SiC, C
^|2568| L + C = B4C + SiC, BN | D2
L + C = B.C + SiC L, SiC,
C + SiC t B.C + BN B,C, BN
L + B,C = B, BN
* B = B4C + SiBn, BN |D3 L * B = Sia, BN
+ SiB» + BN L, BaC,
SiB., BN
2123 | dlT J2123 |L t S i a = SiB., B.CTBN | D<

* SiB. = SiB., B.C
L, SiB,, B,C, BN | 2114 | de

G + L = Si,N«, SiC G + L = Si,N,, BN
757 G * SiC = C + SbN., BN I De |-

3 + SiC = C + Sbt
N,+ BN
1687 I dip
|1687| L = Si, SiiN., BN, SiC |D7 h
| L = Si, SiC, Si,N, L = Si, SbN., BN
L, Si,
BN, SiC
L + SiC = B4C + Si
SiC + B.C + Si + BN
I 1657 I dii 1657 I di2 "

|1657| L = Si + SiB., BN, B.C I Da
L = Si + SiB,, B.C L = Si + SiB., BN
Si, SiB,,
BN, B,C
1471 | di3 1471 I d u 1

i + SiB.=SiB,, B.C i t S i & = S i B , , BN|
Fig. 26. The Scheil reaction scheme of the Si-B-C-N system
Si:B:C ratio with regard to the PHBS(p)-derived material was fixed with
Si34.3B12.4C53,3 and the nitrogen potential was varied.
Along the upper line Si3N4, SiC and graphite are in equilibrium with the gas
phase. When crossing this line, one of the phases disappears according to the
reaction (2). Above the line the solid phases Si3N4, graphite and BN are in
equilibrium with the gas phase whereas below the line graphite, SiC and BN are
in equilibrium with the gas phase. For the given Sia^B^^Css^ composition
some excess graphite remains after the complete consumption of Si3N4.
Increasing the N2 pressure from p N 2 = 1 bar to 10 bar the temperature of
reaction (2) is increased from 1757 to 1973 K which also indicates a
stabilization of Si3N4. The potential phase diagrams for other Si-B-C-N
60
SiC
50-
40-
Si3N4
30- Graphite
SiC
f 20- BN
Gas
Graphite
10-
1000 1500 2000 2500 3000

Temperature [K]
Fig. 27. Phase fraction diagram for Si-B-C-N ceramics derived from PHBS(p) precursor
[273]
- 1 bar
- 0 . 1 bar
1200 1400 1600 1800 2000 2200 2400 2600 2800

Temperature [K]
Fig. 28. Calculated log(p N2 )-temperature diagram (for Si34.3B12.4C53.3 composition) [273]
compositions are shown in [244]. A further effect increasing the temperature

of reaction (2) is due to the dissolution of carbon in the turbostratic B-N-C
layers. Consequently the activity of carbon is decreased. Calculated activity vs.
temperature diagrams quantify this effect on phase equilibria as shown in
Fig. 29 [272, 273].
At a pressure of 1 bar the reaction temperature (2) is increased from 1757 K
at an activity a c = 1 to 2114 K at an activity of a c = 0.1. At lower carbon
activities Si3N4 decomposes according to reaction (1) as indicated by the
horizontal line. At an assumed pressure of 10 bar the reaction temperature (2)
I 1 1
2400- Liquid /
2300- 72307 SiC -
2- 2200 - -
§ 2100- /2114
""""--..._ 10 bar
» 2000 -
£ 1973
P 1900-
1800- ^ ^ ~ — - — i _ _ 1 bar
Si,N4
1757
1700-
0.2 0.4 0.6 0.8 1.0

Activity C
Fig. 29. Temperature-activity (ac) diagram in the Si-C-N system [272, 273]
is 1973 K and increases to 2307 K at an activity a c of 0.17. At lower activities

again reaction (1) occurs. The calculated reaction temperatures are consistent
with values given in Figs. 6, 21 and 28, respectively. It has to be emphasized
that values for the p N 2 and the activities are not known. Layers of separated BN
and graphite also may be possible instead of BCN layers.
Conclusion
Thermodynamic calculations according to the CALPHAD method can be used
to track phase equilibria and phase reactions of ceramics and to contribute
to the understanding of the materials processing. Combinations of such
calculations with selected experiments are the basis for an efficient materials
development. Comparison with available experimental results show that in the
quaternary Si-B-C-N system the calculations are reliable. Based on a consistent
thermodynamic dataset the most practically useful types of phase diagrams
(isothermal sections, temperature-composition diagrams), phase composition
diagrams and property diagrams (e.g. phase fraction diagrams) can be
calculated. The results are valid to find guidelines for favorable sintering and
processing conditions and the understanding of the materials high-temper-
ature reactions during application.
However, to use thermodynamically calculated results for the interpretation
requires the consideration of the materials microstructure, specific crystalli-
zation behavior and the kinetics of phase formation and phase reaction. This is
true especially for the interpretation of dynamic thermal analysis experiments
(e.g. DTTA, TG). Moreover, heat treatments (e.g. thermolysis) are often carried
out in an inert atmosphere and evaporating gaseous species are continuously
removed by flowing gas atmospheres (inert or reactive). This effect deeply
influences the phase reactions. Therefore, thermodynamic calculations for the

simulation of such processes are in progress.
Acknowledgements. This work was financially supported by the Deutsche Forschungs-

gemeinschaft (DFG). We thank H.L. Lukas and J. Peng for detailed discussions and
M. Thomas for support in preparation of the manuscript.
8
References
1. Riedel R (ed) (2000) Handbook of ceramic hard materials, Vol 1 and Vol 2. Wiley-
VCH, Weinheim
2. Petzow G, Hermann M (2002) Silicon Nitride Ceramics. In: Jansen MA (ed) Structure
and Bonding, Vol. 102. Springer, Berlin, Heidelberg, New York
3. German RM (1985) Liquid phase sintering, Plenum Press, New York
4. Telle R (1990) Structure and properties of Si-doped boron carbide. In: Freer R (ed) The
Physics and Chemistry of Carbides, Nitrides and Borides. Kluwer Academic Publishers,
Dordrecht, p 249
5. Baldus HP, Wagner O, Jansen M (1992) Mat Res Soc Symp Proc 271: 821
6. Bill J, Aldinger F (1995) Adv Mater 7: 775
7. Riedel R, Bill J, Kienzle A (1996) Appl Metallorg Chem 10: 241
8. Jansen M, Jiingermann H (1997) Curr Op Solid State & Mater Sci 2: 150
9. (a) Baldus HP, Jansen M (1997) Angew Chem 109: 338; (b) Baldus HP, Jansen M (1997)
Angew Chem Int Ed Engl 36: 328
10. Laine RM, Sellinger A (1998) Si-containing ceramic precursors. In: Rappoport Z,
Apeloig Y (eds) The Chemistry of Organic Silicon Compounds. Vol. 2, J. Wiley & Sons
Ltd., London, p 2245
11. Baldus HP, Jansen M, Sporn D (1999) Science 285: 699
12. Bill J, Wakai F, Aldinger F (eds) (1999) Precursor-derived ceramics. Wiley-VCH,
Weinheim, New York, 36: 328
13. Weinmann M, Schuhmacher J, Kummer H, Prinz S, Peng J, Seifert HJ, Christ M, Miiller
K, Bill J, Aldinger F (2000) Chem Mater 12: 623
14. Filipozzi L, Derre A, Conard J, Piraux L, Marchand A (1995) Carbon 33: 1747
15. Hegemann D, Riedel R, Oehr C (1999) Chem Vap Deposition 5: 61
16. Suenaga K, Willaime F, Loiseau A, Colliex C (1999) Appl Phys A 68: 301
17. Pampuch R (1999) J Europ Ceram Soc 19: 2395
18. Bill J, Schuhmacher J, Miiller K, Schempp S, Seitz J, Diirr J, Lamparter HP, Weinmann
M, Golczewski J, Peng J, Seifert HJ, Aldinger F (2000) Z Metallkd 91: 335
19. Wiedemeier H, Singh M (1989) Interfacial characterization of silicon nitride reinforced
ceramic-matrix composites: a thermodynamic approach. In: Lin RY, Arsenault RJ,
Martins GP, Fishman SG (eds) Interfaces in Metal-Ceramics Composites. The Minerals,
Metals & Materials Society
20. Lee KN, Jacobson NS (1995) J Am Ceram Soc 78: 711
21. Riedel R, Greiner A, Miehe G, Dreftler W, Fuefi H, Bill J, Aldinger F (1997) Angew
Chem 109: 657
22. Liu AY, Cohen ML (1989) Science 245: 841
23. Kroll P, Hoffmann R (1998) Angew Chem Int Ed 37: 2527
24. Saunders N, Miodownik P (1988) CALPHAD (Calculation of phase diagrams): A
comprehensive guide. In: Cahn RW (ed) Pergamon Materials Series, Vol 1, Pergamon,
Oxford
25. Lukas HL, Henig ETH, Petzow G (1986) Z Metallkd 77: 360
26. Villars P, Calvert LD (1991) Pearson's handbook of crystallographic data for
intermetallic phases, 2 nd edn, ASM, Metals Park, Ohio
27. Viala JC, Bouix J (1980) J Less-Common Met 71: 195

28. Zerr A, Miehe G, Serghiou G, Schwarz M, Kroke E, Riedel R, Fuefi H, Kroll P, Boehler R
(1999) Nature 400: 340
29. Lukas HL, Henig ET, Zimmermann B (1977) CALPHAD 1: 225
30. Lukas HL, Fries SG (1992) J Phase Equilibria 13: 1261
31. Dinsdale AT (1991) CALPHAD 15: 319
32. Scientific Group Thermodata Europe, Grenoble Campus, 1001 Avenue Centrale, BP66,
F-38402 Saint Martin D'Heres, France
33. Kasper B (1996) PhD Thesis, Universitat Stuttgart
34. Kasper B, Seifert HJ, Kufimaul A, Lukas HL, Aldinger F (1997) Entwicklung eines
thermodynamischen Datensatzes fur das System B-C-N-Si-O. In: Aldinger F, Mughrabi
H (eds) Proc Werkstoffwoche '96, Stuttgart, Symp 7, Materialwissenschaftliche
Grundlagen. DGM-Informationsgesellschaft mbH, Frankfurt, p 623
35. Grobner J, Lukas HL, Aldinger F (1996) CALPHAD 20: 247
36. Lim SK, Lukas HL (1995) Thermodynamische Optimierung des Systems B-C-Si und
seiner Randsysteme. In: Petzow G, Tobolski J, Telle R (eds) Hochleistungskeramik,
Herstellung, Aufbau und Eigenschaften, Deutsche Forschungsgemeinschaft. VCH,
Weinheim, p 605
37. Hillert M, Jonsson S, Sundman B (1992) Z Metallkd 83: 648
38. Liang JJ, Topor L, Navrotsky A, Mitomo A (1999) J Mater Res 14: 1959
39. Redlich O, Kister AT (1948) Ind Eng Chem 40: 345
40. Muggianu YM, Gambina M, Bros JP (1975) J Chim Phys 72: 83-88
41. Hillert M (2001) Journal Alloys and Compounds 320: 161
42. Moissan H, Stock A (1900) C R Acad Sci (Paris) 131: 139
43. Knarr WA (1959) PhD thesis, University of Kansas
44. Brosset C, Magnusson B (1960) Nature 187: 54
45. Trumbore FA (1960) Bell Syst Tech J 39: 205
46. Samsonov GV, Sleptsov VM (1963) Russ J Inorg Chem 8: 1047
47. Hesse J (1968) Z Metallkd 59: 499
48. Vick GL, Whittle KM (1969) J Electrochem Soc 116: 1142
49. Ettmayer P, Horn HC, Schwetz KA (1970) Mikrochim Acta Suppl IV: 87
50. Schwettmann FN (1974) J Appl Phys 45: 1918
51. Armigliato A et al and Masetti G et al (1977) In: Huff HR, Sirtl E (eds) Semiconductor
silicon 1977. Electrochem Soc Inc, Princeton, New York, p 638 and p 648.
52. Arabei BG (1979) Izv Akad Nauk SSSR, Neorg Mat 15: 1589
53. Lugscheider E, Reimann H, Quadakkers WJ (1979) Ber Dt Keram Ges 56: 301
54. Ryssel H, Miller K, Haberger K, Henkelmann R, Jahnel F (1980) Appl Phys 22: 35
55. Viala JC, Hillel R, Bouix J (1980) J Less-Common Met 71: 207
56. Armas B, Male G, Salanoubat D, Chatillon C, Allibert M (1981) J Less-Common Met 82:
245
57. Male G, Salanoubat D (1981) Rev Int Hautes Temp Fr 18: 109
58. Vlasse N, Slack GA, Garbauskas M, Kasper JS, Viala JC (1986) J Sol State Chem 63: 31
59. Borisenko VE, Yudin SG (1987) Phys Stat Sol (A) 101: 123
60. Bolmgren H (1994) On the crystal structure and homogeneity range of rhombohedral
silicon boride. Report UUIC-B18-266. Institute of Chemistry, Uppsala University,
Sweden
61. Aselage TL (1998) J Mater Res 13: 1786
62. Olesinski RW, Abbaschian GJ (1984) Bull Alloy Phase Diagrams 5: 478
63. Giese RF, Economy J, Matkovich VI (1965) Z Krist 122: 144
64. Magnusson B, Brosset C (1962) Acta Chem Scand 16: 449
65. Bolgar AS, Blinder AV, Muratov VB, Makarenko GN (1990) Thermodynamic properties
of boron silicides. In: Silicides and their application in technology: Collection of
scientific papers (in Russian). Inst Probl Materialoved, Akad Nauk UkrSSR, Kiev, p 78
66. Esin YuO, Kolesnikov SP, Baev VM, Ermakov AF (1978) Enthalpies of formation of
liquid alloys of silicon with boron (in Russian). Tezisy Nauchn Soobshch Vses Konf Str
Svoistvam Met Shlakovykh Rasplavov, 3 r d edn., Vol 2, p 182
67. Biletskii AK, Scheretskii AA, Vitusevich VT, Shumihin VT (1988) Metals 3: 66
68. Gordienko SP (1995) Powder Metallurgy and Metal Ceramics 34: 660
69. Armas B, Chatillon C, Allibert M (1981) Rev Int Hautes Temper Refract 18: 153
70. Noguchi R, Suzuki K, Tsukihashi F, Sano N (1994) Met Mater Trans B 25B: 903
71. Tanahashi M, Fujisawa T, Yamauchi C (1998) Shigen-to-Sozai 114: 807
72. Zaitsev AI, Kodentsov AA (2001) J Phase Equilibria 22: 126
73. Kaufman L, Uhrenius B, Birnie D, Taylor K (1984) CALPHAD 8: 25
74. Dirx RR, Spear KE (1987) CALPHAD 11: 167
75. Goujard S, Vandenbulcke L, Bernard C (1994) CALPHAD 18: 369
76. Goujard S, Vandenbulcke L, Bernard C, Blondiaux G, Debrun JL (1994) J Electrochem
Soc 141: 452
77. Kistler-de Coppi (1985) PhD Thesis, ETH Zurich
78. Hall RN (1958) J Appl Phys 29: 914
79. Scace RI, Slack GA (1959) J Chem Phys 30: 1551
80. Dolloff RT (1960) Research study to determine the phase equilibrium relations of
selected metal carbides at high temperatures. WADD Technical Report 60-143, Wright
Air Development Division
81. Nozaki eT, Yatsurugi Y, Akiyama N (1970) J Electrochem Soc 117: 1566
82. Voltmer FW, Padovani FA (1973) Semicond Silicon, Pop Int Symp Silicon Mater Sci
Technol 2 nd : 75-82
83. Oden LL, McCune RA (1987) Met Trans A 18 A: 2005
84. Suhara S, Yuge N, Fukai M, Aratani F (1989) CAMP-ISIJ 2: 1341
85. Kleykamp H, Schumacher G (1993) Ber Bunsenges Phys Chem 97: 799
86. Yanaba K, Akasaka M, Takeuchi M, Watanabe M, Narushima T, Iguchi Y (1997) Mater
Trans JIM 38: 990
87. Ottem (1997) cited in [86]
88. Olesinski RW, Abbaschian GJ (1984) Bull Alloy Phase Diagrams 5: 486
89. Kieffer R, Gugel E, Ettmayer P, Schmidt A (1969) Mater Res Bull 4: 153
90. Humphrey GL, Todd SS, Coughlin JP, King EG (1952) Report RI4888. US Dept Interior,
Bureau of Mines
91. Kelley KK (1941) J Am Chem Soc 63: 1137
92. Walker BE, Ewing CT, Miller RR (1962) J Chem Eng Data 7: 595
93. Koshchenko VI, Demidenko AF, Prokofev NK, Yachmenev VE, Radchenko AF (1979)
Inorg Mater 15: 1017
94. Koshchenko VI, Grinberg YK, Koshchenko RV (1985) Inorg Mater 21: 197
95. Magnus A (1923) Ann d Physik IV 70: 303
96. Guseva EA, Bolgar AS, Gorbaenko SP, Gorbatyuk VA, Fesenko VV (1966) Teplofiz
Vyschikh Temp 4: 649
97. Chekhovskoy VY (1971) J Chem Thermodyn 3: 289
98. Ruff O, Grieger P (1933) Z Anorg Allg Chem 211: 145
99. Yudin BF, Markholiya TP (1969) Russ J Appl Chem 42: 2439
100. Greenberg E, Natke CA, Hubbard WN (1970) J Chem Thermodyn 2: 193
101. Chipman J, Fulton JC, Gokcen N, Caskey Jr GR (1954) Acta Metallurg 2: 439
102. Drowart J, De Maria G, Inghram MG (1958) J Chem Phys 29: 1015
103. Grieveson P, Alcock CB (1960) The thermodynamics of metal silicides and silicon
carbide. In: Popper P (ed) Special Ceramics. Heywood, London, p 183
104. Davis SG, Anthrop DF, Searcy AW (1961) J Chem Phys 34: 659
105. Kirkwood DH, Chipman J (1961) J Phys Chem 65: 1082
106. d'Entremont JC, Chipman J (1963) J Phys Chem 67: 499
107. Rein RH, Chipman J (1963) J Phys Chem 67: 839
108. Sambasivan S, Capobianco C, Petuskey W (1993) J Am Ceram Soc 76: 397
109. (a) Rocabois P, Chatillon C, Bernard C (1995/96) High Temperatures-High Pressures

27/28: 3; (b) Rocabois P, Chatillon C, Bernard C (1995/96) High Temperatures-High
Pressures 27/28: 25
110. Kleykamp H (1998) Ber Bunsengesellschaft Phys Chem 102: 1231
111. Coltters RG (1985) Mater Sci Eng 76: 1
112. Gurvich LV, Veyts IV, Alcock CB (eds) (1991) Thermodynamic properties of individual
substances. Fourth edition, Vol 1 and 2, Hemisphere Publishing Corporation
113. Kaufman L (1979) CALPHAD 3: 45
114. Carlson ON (1990) Bull Alloy Phase Diagrams 11: 569
115. Kaufman L (1979) CALPHAD 3: 275
116. Dorner P, Gauckler LJ, Krieg H, Lukas HL, Petzow G, Weiss J (1981) J Mater Sci 16: 935
117. Weiss J, Lukas HL, Lorenz J, Petzow G, Krieg H (1981) CALPHAD 5: 125
118. Kaiser W, Thurmond CD (1959) J Appl Phys 30: 427
119. Yatsurugi Y, Akiyama N, Endo Y (1973) J Electrochem Soc 120: 975
120. Narushima T, Ueda N, Takeuchi M, Ishii F, Iguchi Y (1994) Mater Trans JIM 35: 821
121. Kostanovskii AV, Evseev AV (1994) High Temperature 32: 25
122. Cerenius Y (1999) J Am Ceram Soc 82: 380
123. Guzman IYa, Demidenko AF, Koshchenko VI, Fraifel'd MS, Egner YuV (1976) Inorg
Mater 12: 1546
124. Koshchenko VI, Grinberg Y (1982) Inorg Mater 18: 903
125. Rocabois P, Chatillon C, Bernard C (1996) J Am Ceram Soc 79: 1351-1360
126. Neel DC, Pears CD, Oglesby S (1962) The thermal properties of thirteen solid materials
to 5000 °F or their destruction temperatures. Tech Doc Rep WADD-TR-60-924. South
Res Inst, Birmingham
127. Wood JL, Adams GP, Mukerji J, Margrave JL (1973) Proc Third Int Conf Chem
Thermodyn, Baden bei Wien, 2: 115
128. (a) O'Hare PAG, Tomaszkiewicz I, Beck II CM, Seifert HJ (1999) J Chem Thermodyn
31: 303; (b) O'Hare PAG, Tomaszkiewicz I, Seifert HJ (1997) J Mater Res 12: 3203
129. Matignon C (1913) Bull Soc Chim Fr 13: 791
130. Hincke WB, Brantley LR (1930) J Am Chem Soc 52: 48
131. Pehlke RD, Elliott JF (1959) Trans Metall Soc AIME 215: 781
132. McLaine LA, Coppel CP, Little AD (1966) Equilibrium studies of refractory nitrides.
Technical Report AFML-TR-65-99, Part 11, Dec 1966
133. Ryklis EA, Bolgar AS, Fesenko VV (1969) Poroshk Metall (Kiev) 73: 92
134. Colquhoun I, Wild S, Grieveson P, Jack KH (1973) Proc Brit Ceram Soc 22: 207
135. Wild S, Grieveson P, Jack KH (1972) The thermodynamics and kinetics of formation of
phases in the Ge-N-0 and Si-N-0 systems. In: Popper P (ed) Special Ceramics 5. British
Ceramic Research Association, Stoke-on-Trent, Chap 20, p 271
136. Blegen K (1977) Equilibria and kinetics in the systems Si-N and Si-N-O. In: Riley FL
(ed) Nitrogen Ceramics. Proc Nato Advanced Study Institute on Nitrogen Ceramics
Noordhoff, Leyden, Chap 21, p 223
137. Andrievskii RA, Lyutikov RA (1996) Russ J Phys Chem 70: 526
138. Durham SJP, Shanker K, Drew RAL (1991) Canadian Metallurgical Quarterly 30: 39
139. Hendry A (1977) Thermodynamics of silicon nitride and oxynitride. In: Riley FL (ed)
Nitrogen Ceramics. Proc Nato Advanced Study Institute on Nitrogen Ceramics,
Noordhoff, Leyden, p 183
140. Chase Jr. MW, Davis CA, Downey Jr. JR, Frurip DJ, McDonald RA, Syverud AN (1985)
JANAF Thermochemical Tables. 3 r d edn, J Phys Chem Ref Data 14. American Chemical
Soc, American Inst of Physics, Nat Bureau of Standards
141. Heuer AH, Lou VLK (1990) J Am Ceram Soc 73: 2785
142. Elliott RP, Van Thyne RJ (1961) The boron-carbon system. Final Technical Report ARF
2200-12, US Atomic Energy Commission
143. Lowell CE (1967) J Am Ceram Soc 50: 142
144. Kieffer R, Gugel E, Leimer G, Ettmayer P (1971) Ber Dt Keram Ges 48: 385
145. (a) Bouchacourt M, Thevenot F (1981) J Less-Common Met 82: 219; (b) Bouchacourt
M, Thevenot F (1981) J Less-Common Met 82: 227
146. Beauvy M (1983) J Less-Common Met 90: 169
147. Samsonov GV, Zhuravlev NN, Annuel IG (1956) Fiz Met Metalloved 3: 309
148. Zhuravlev NN, Makarenko G, Samosonov GV (1960) Izvest Akad Nauk SSSR, Otdel
Tekh Nauk Met I Toplivo 1: 133
149. Bouchacourt M, Thevenot F, Ruste J (1978) J Less-Common Met 59: 139; Bouchacourt
M, Thevenot F (1979) J Less-Common Met 67: 327
150. Thevenot F (1990) J Europ Ceram Soc 6: 205
151. Schwetz KA, Karduck P (1991) J Less-Common Met 175: 1
152. (a) Werheit H, Kuhlmann U, Laux M, Lundstrom T (1993) Phys Stat Sol B 179: 489;
(b) Werheit H, Kuhlmann U, Lundstrom T (1994) J Alloys Compounds 204: 197
153. Okamoto H, J Phase Equilibria 13: 436
154. Gilchrist KE (1985) High Temp-High Pressures 17: 671
155. Wood C, Emin D, Gray PE (1985) Phys Rev B (Solid State) 31: 6811
156. Matsui T, Arita Y, Naito K, Imai H (1991) J Nucl Mater 186: 7
157. Medwick PA, Fischer HE, Pohl RO (1994) J Alloys Compounds 203: 67
158. King EG (1949) Ind Eng Chem 41: 1298
159. Sheindlin AE, Belevich IS, Kozhenikov IG (1972) Teplofiz Vysok Temp 10: 421
160. Smith D, Dworkin AS, Van Artsdalen ER (1955) J Am Chem Soc 77: 2654
161. Domalski ES, Armstrong GT (1968) J Res Nat Bureau Std-A, Phys and Chem 72A: 133
162. Gal'chenko GL, Timofeev BI, Makarenko GN, Samsonov GV (1970) Russ J Phys Chem
44: 1400
163. Hong KC, Kleppa OJ (1978) J Chem Thermodyn 10: 797
164. Froment K, Fouletier J, Fouletier M (1991) J Appl Electrochem 21: 175
165. Froment AK, Chatillon C, Colin M (1991) Rev Int Htes Temp Refract 27: 141
166. Froment K, Chatillon C, Fouletier J, Fouletier M (1992) J Nucl Mater 188: 280
167. Dorner P (1982) PhD thesis, Universitat Stuttgart
168. Kuhlmann U, Werheit H, Schwetz KA (1992) J Alloys Comp 189: 249
169. Seifert HJ, Kasper B, Lukas HL, Aldinger F (2001) Thermodynamic assessment of the
B-C system. In: Marin-Ayral RM, Record MC (eds) Compte-Rendus, Journees d'Etude
des Equilibres entre Phases (XXVII JEEP). March 22 and 23, 2001, Montpellier, France,
p 123
170. (a) Ploog K, Rauh P, Stoeger W, Schmidt H (1972) J Cryst Growth 13/14: 350; (b) Ploog
K, Schmidt H, Amberger E, Will G, Kossobutzki KH (1972) J Less-Common Met 29: 161
171. Saitoh H, Yoshida K, Yarbrough WA (1993) J Mater Res 8: 8
172. Wen H (1994 ) Thermodynamische Berechnungen der Konstitution des Systemes
Al-B-C-N-Si-Ti. VDI Fortschrittsberichte, Reihe 5, Nr. 333, Diisseldorf
173. Bundy FP, Wentorf RH (1963) J Chem Phys 38: 1144
174. Corrigan FR, Bundy FP (1975) J Chem Phys 63: 3812
175. Solozhenko VL, Turkevich VZ, Holzapfel WB (1999) J Phys Chem B 103: 2903
176. Solozhenko VL (1988) Dokl Phys Chem 301: 592
177. Dvorkin AS, Sasmor DJ, Van Artsdalen ER (1954) J Chem Phys 22: 837
178 Sirota NN, Kofman NA, Petrova ZhK (1975) Izv. AN BSSR, Ser Fiz-Mat Nauk (USSR) 6:
75
179. Gorbunov VE, Gavrichev KS, Totrova GA, Bochko AV, Lazarev VB (1988) Russ J Phys
Chem 62: 9
180. Gavrichev KS, Solozhenko VL, Gorbunov VE, Golushina LN, Totrova GA, Lazarev VB
(1993) Thermochim Acta 217: 77
181. Sirota NN, Kofman NA (1975) Sov Phys-Dokl 20: 861
182. Solozhenko VL, Yachmenev VE, Vil'kovsky VA, Sokolov AN, Shul'zhenko AA (1987)
Russ J Phys Chem 61: 1480
183. Gorbunov VE, Gavrichev KS, Totrova GA, Bochko AV, Lazarev VB (1987) Russ J Phys
Chem 61: 3357
184. Solozhenko VL, Yachmenev VE, Vil'kovsky VA, Petrusha IA (1989) Inorg Mater 25: 134
56 H.J, Seifert • P. Aldinger
185. Atake T, Honda A, Saito Y, Saito K (1990) Jpn J Appl Phys 29: 1869
186. Sirota NN, Kofman NA (1976) Sov Phys-Dokl 21: 516
187. Gavrichev KS, Gorbunov VE, Solozhenko VL, Totrova GA, Golushina LN (1992) Zh Fiz
Khim 66: 2824
188. Prophet H, Stull DR (1963) J Chem Eng Dat 8: 78
189. Lusternik VE, Solozhenko VL (1992) Zh Fiz Khim 66: 1186
190. McDonalds RA, Stull DR (1961) J Phys Chem 65: 1918
191. Agoshkov VM, Bogdanova SV (1990) Sov J Superhard Mater 12: 26
192. Solozhenko VL (1993) Zh Fiz Khim 67: 1580
193. Mezaki R, Tilleux EW, Barnes DW, Margrave JL (1962) In: Thermodynamics of Nuclear
Materials. IAEA, Vienna, p 494
194. Kiseleva IA, Mel'chakova LV, Topor ND (1963) Izv Akad Nauk SSSR Neorg Mater
(USSR) 9: 256
195. Solozhenko VL, Chaikovskaya IYA, Sokolov AN, Shul'zhenko AA (1987) Russ J Phys
Chem 61: 412
196. Solozhenko VL, Chaikovskaya IYa, Petrusha IA (1989) Inorg Mater 25: 1414
197. Mayorova AF, Mudretsova SN, Solozhenko VL, Ryabova LA (1992) Proc Int Ceramic
Congress, Istanbul, Turkey, 19-23 Oct 1992 (TCS, Istanbul, 1992) p 659
198. Solozhenko VL (1993) Thermochim Acta 218: 221
199. Solozhenko VL (1993) Thermochim Acta 218: 395
200. Solozhenko VL, Turkevich VZ (1992) J Therm Anal 38: 1181
201. Solozhenko VL (1995) J Therm Anal 44: 97
202. GaPchenko GL, Kornilov AN, Skuratov SM (1960) Russ J Inorg Chem 5: 1282
203. Wise SS, Margrave JL, Feder HM, Hubbard WN (1966) J Phys Chem 70: 7
204. Leonidov VYa, Tirnofeev IV (1989) Russ J Inorg Chem 34: 1545
205. Leonidov VYa, Timofeev IV, Solozhenko VL, Rodionov IV (1987) Russ J Phys Chem 61:
1503
206. Leonidov VYA, Timofeev IV, Lazarev VB, Bochko AV (1988) Zh Neorg Khim (USSR)
33: 1597
207. Leonidov VYa, O'Hare PAG (1992) Pure Appl Chem 64: 103
208. Hoch M, White D (1956) Technical Research Report MCC 1023 TR 214. Ohio State
University Research Foundation, Columbus, Ohio
209. Schissel P, Williams W (1959) Bull Am Phys Soc 4: 139
210. Dreger LH, Dadape VV, Margrave JL (1962) J Phys Chem 66: 1556
211. Hildenbrand DL, Hall WF (1963) J Phys Chem 67: 888
212. Gavrichev KS, Solozhenko VL, Lazarev VB (1994) Inorg Mater 30: 1025
213. Solozhenko VL (1995) J Hard Mater 6: 51
214. Rogl P (2001) Int J Inorg Mater 3: 201
215. (a) Goenna J, Meurer HJ, Nover G, Peun T, Schoenbohm D, Will G (1998) Mater Lett
33: 321; (b) Will G, Nover G, Goenna J (2000) J Solid State Chem 154: 280
216. Prochazka S (1981) The sintering process for SiC, a review. Technical report
81-CRD-314, General Electric, Schenectady, NY
217. Portnoi KI, Samsonov GV, Solonnikova LA (1960) Russ J Inorg Chem 5: 988
218. Secrist DR (1964) J Am Ceram Soc 47: 127
219. Meerson GA (1966) Investigations of some hard alloys in the boron-silicon-carbon
system. Mod Develop Powder Met, Proc Int Powder Met Conf 3: 95
220. Dokukina IV, Kalinina AA, Sokhor MI, Shamrai FI (1967) Izv Akad Nauk SSSR, Neorg
Mater 3: 630
221. Shaffer PTB (1969) Mat Res Bull 4: 213
222. Kalinina AA, Sokhor MI, Shamrai FI (1971) Izv Akad Nauk SSSR, Neorg Mat 7: 778
223. (a) Gugel E, Kieffer R, Leimer G, Ettmayer P (1972) Investigation in the ternary system
boron-carbon-silicon. Nat Bur Stand Spec Publ, Solid State Chemistry, Proc 5 Mat
Res Symp 364: 505; (b) Kieffer R, Gugel E, Leimer G, Ettmayer P (1972) Ber Dt Keram
Ges 49: 41
224. Gierlotka S, Oleksyn O, Palosz B (1994) Mater Sci Forum 166-169: 529
225. Werheit H, Kuhlmann U, Laux M, Telle R (1994) J Alloys and Compounds 209: 181
226. Lipp A, R6der M (1966) Z Anorg Allg Chemie 344: 225
227. (a) Kato T, Yoshimura M, Somiya S (1981) Yogyo-Kyokai-Shi 89: 356; (b) Kato T,
Yoshimura M, Somiya S (1986) Rep Res Lab Eng Mat, Tokyo Inst Tech 11: 45
228. Isomura K, Fukuda T, Ogasahara T, Funahashi T, Uchimura R (1989) Proc UNITECR
89. Trosel LS (ed) J Am Ceram Soc 624
229. Thevenot F, Doche C, Mongeot H, Guilhon F, Miele P, Cornu D, Bonnetot B (1997)
J Solid State Chem 133: 164
230. Baldus HP, Jansen M, Wagner O (1994) Key Eng Mater 89-91: 75
231. Kavecky S, Janekova B, Sajgalik P (2000) Key Eng Mater 175: 49
232. Kroke E, Li YL, Konetschny C, Lecomte E, Fasel C, Riedel R (2000) Mater Sci Eng
R26: 97
233. Kleebe HJ, Suttor D, Miiller H, Ziegler G (1998) J Am Ceram Soc 81: 2971
234. Seifert HJ, Peng J, Aldinger F (2001) J Alloys and Compounds 320: 251
235. Rassaerts H, Schmidt A (1966) Planseeberichte fur Pulvermetallurgie 14: 110
236. Seifert HJ, Aldinger F (1999) Thermodynamic calculations in the system Si-B-C-N-O.
In: Bill J, Wakai F, Aldinger F (eds) Precursor-Derived Ceramics. Wiley-VCH,
Weinheim, New York, p 165
237. Peng J, Seifert HJ, Aldinger F (2000) Thermal analysis of Si-C-N ceramics derived from
polysilazanes. In: Muller G (ed) Ceramics-Processing, Reliability, Tribology and Wear.
Proc EUROMAT '99, Vol 12, Wiley-VCH, Weinheim, p 120
238. Lorenz J, Hucke EE, Lukas HL, Petzow G (1983) Formation of SiC by the reaction of
Si-containing vapour with porous glassy carbon. In: Vincenzini P (ed) Ceramic
Powders. Elsevier, Scientific Publishing Company, Amsterdam
239. Nickel KG, Hoffmann MJ, Greil P, Petzow G (1988) Adv Ceramic Materials 3: 557
240. Wada H, Wang MJ, Tien TY (1988) J Am Ceram Soc 71: 837
241. Misra AK (1991) J Mater Sci 26: 6591
242. Neidhardt U, Schubert H, Bischoff E, Petzow G (1994) Key Eng Mater 89-91: 187
243. Li Y, Hu Z, Gao Y, Liang Y (1995) J Mater Sci Technol 11: 466
244. Seifert HJ, Lukas HL, Aldinger F (1998) Ber Bunsenges Phys Chem 9: 1309
245. Jha A (1993) J Mater Sci 28: 3069
246. Gugel E, Ettmayer P, Schmidt A (1968) Ber Dt Keram Ges 45: 395
247. Kandori T, Kamiya N, Kamigaito O (1975) Japan J Appl Phys 14: 137
248 Schroder F (ed), Gmelin handbook of inorganic chemistry, Silicon B3 (1986) 8th edn.,
Springer-Verlag Berlin, Heidelberg, New York, Tokyo, p 527
249. Rocabois P, Chatillon C, Bernard C (1999) High Temp-High Press 31: 433
250. J. Seitz (1996) PhD thesis, University of Stuttgart, Germany
251. Seitz J, Bill J (1996) J Mater Sci Lett 15: 391
252. Seifert HJ (1999) Z Metallkd 90: 1016
253. Bill J, Seitz J, Thurn G, Duerr J, Canel J, Janos BZ, Jalowiecki A, Sauter D, Schempp S,
Lamparter HP, Mayer J, Aldinger F (1998) Phys Stat Sol A 166: 269
254. Inomata Y, Inoue Z (1973) Yogyo-Kyokai-Shi 81: 441
255. Riedel R (1994) Adv Mater 6: 549
256. Oscroft RJ, Thompson DP (1991) High-temperature reactions in the B-Al-Si-N-C
system. 4* Int Symp Ceram Mater and Components f Engines. Swedish Ceramic Soc,
p 172
257. Ruh R, Kearns M, Zangvil A, Xu Y (1992) J Am Ceram Soc 75: 864
258. Badzian AR, Niemyski T, Appenheimer S, Olkuoenik E (1972) Graphite-boron nitride
solid solutions by chemical vapor deposition. In: Proc 3 r d Int Conf on Chemical Vapor
Dep, Salt Lake City, UT, p 747
259. Badzian AR (1981) Mater Res Bull 16: 1385
260. Chen SH, Diefendorf RJ (1980) Electrical properties of the system: boron-nitrogen-
carbon. In: Proc Yd Int Carbon Conf Baden-Baden, p 44
261. Kaner RB, Kouvetakis J, Warble CE, Sattler ML, Bartlett N (1987) Mater Res Bull 22: 399
262. Bill J, Riedel R, Passing G (1992) Z Anorg Allg Chem 610: 83
263. Saugnac F, Teyssandier F, Marchand A (1992) J Am Ceram Soc 75: 161

264. Derre A, Filipozzi L, Bouyer F, Marchand A (1994) J Mater Sci 29: 1589
265. Sirota NN, Zhuk MM, Mazurenko AM, Olekhnovich AI (1977) Vestsi Akad Nauk BSSR,
Ser Fiz-Mat Nauk 2: 111; Chemical Abstracts (1977) 87: 59009
266. Sauter DH (2000) PhD thesis, Universitat Stuttgart, Germany
267. Solozhenko VL, Turkevich VZ, Sato T (1997) J Am Ceram Soc 80: 3229
Solozhenko VL (1997) Europ J Solid State Inorg Chem 34: 797
268. Nicolich JP, Hofer F, Brey G, Riedel R (2001) J Am Ceram Soc 84: 279
269. Solozhenko VL, Andrault D, Fiquet G, Mezouar M, Rubie DC (2001) Appl Phys Lett 78:
1385
270. Bill J, Kamphowe TW, Miiller A, Wichmann T, Zern A, Weinmann M, Schuhmacher J,
Miiller K, Peng. J, Seifert HJ, Aldinger F (2001) Appl Organomet Chem 15: 777
271. Seifert HJ, Peng J, Aldinger F (1999) Die Konstitution von Si-B-C-N Keramiken. In:
Heinrich J, Ziegler G, Hermel W, Riedel R (eds) Werkstoffwoche '98, Vol VII, Keramik/
Simulation Keramik. Wiley-VCH, Weinheim, p 339
272. Peng J, Seifert HJ, Aldinger F (2000) Thermal stability of precursor-derived Si-(B-)C-N
ceramics. In: Bansal NP, Singh JP (eds) Innovative Processing and Synthesis: Ceramics,
Glasses and Composites IV. Ceramic Transactions, Vol 115, The American Ceramic
Society, USA, p 251
273. Seifert HJ, Peng J, Aldinger F (2001) Appl Organomet Chem 15: 794
274. Singh RN, Brun MK (1988) Adv Ceram Mater 3: 235
275. Jalowiecki A, Bill J, Aldinger F, Mayer J (1996) Composites Part A 27 A: 717
Silicon Carbide - A Survey of Synthetic Approaches,
Properties and Applications
G. Roewer1'*, U. Herzog*, K. Trommer*, E. Miiller2'**, S. Friihauf**
* Institut fur Anorganische Chemie, TU Bergakademie Freiberg, 09596 Freiberg,
Germany
** Institut fur Keramische Werkstoffe, TU Bergakademie Freiberg, 09596 Freiberg,
Germany
1
e-mail: gerhard.roewer@chemie.tu-freiberg.de
2
e-mail: mueller@ikw.tu-freiberg.de
The challenge to develop tailor-made precursor molecules for silicon carbide has
significantly intensified the progress in synthesis of organosilicon polymers. A critical
overview is presented regarding useful synthesis routes towards such molecules with
appropriate chemical composition as well as controllable architecture (metal condensation
of halogenosilanes or silahalogenocarbons, disproportionation of disilanes, dehydrocou-
pling of hydrosilanes, hydrosilylation of olefins).
Furthermore, results are summarized concerning the pathways of the pyrolytic chemical
degradation of the organosilicon precursor including the structure rearrangements in the
solid state.
Finally some illustrative examples of the production of net-shaped ceramic bodies are
demonstrated.
Keywords: Polysilanes, Polycarbosilanes, Dehydrocoupling, Disproportionation, Pyrolysis,

Net-shaped ceramics
1 Introduction 60
2 Synthesis of Precursor Molecules 61

2.1 Precursor-Concept 61
2.2 Type of Precursors and Synthesis Principles 63
2.3 SiC Synthesis from Small Precursor Molecules 63
2.4 Polysilanes as Synthons Towards Silicon Carbide 66
2.4.1 Polysilanes via Wurtz-Coupling 66
2.4.2 Disproportionation of Disilanes 69
2.4.3 Polysilanes via Dehydro-Coupling 72
2.4.4 Other Synthetic Routes to Polysilanes 75
2.5 Polycarbosilanes 76
2.5.1 Polycarbosilanes via Grignard Coupling 77
2.5.2 Polycarbosilanes via ROP of Sila- and Disilacyclobutanes 79
2.5.3 Polycarbosilanes by Hydrosilylation 84
3 Pyrolysis 85
3.1 Conversion of Organosilicon Precursor to Inorganic Material . . 85
3.1.1 Polysilanes/Carbosilanes 86

60 G. Roewer • U. Herzog • K. Trommer • E. Miiller • S. Friihauf
3.1.2 Pyrolysis under Ammonia 90

3.1.3 Poly(chloromethylsilanes/carbosilanes) 91
3.2 Transformation of the Amorphous Si-C Network
into Polycrystalline SiC 92
3.2.1 Amorphous Covalent Ceramics 92
3.2.2 Nucleation and Crystal Growth 95
4 Alternative Routes to SiC 103
4.1 Routes Using Phytogenic Material 103
4.1.1 Routes Starting from Sugar 103
4.1.2 Routes Starting from Silica-Containing Plants 104
4.1.3 Routes Starting from Wood 105
4.2 Silicon Carbide via Gas Phase Processes 106
4.2.1 Single Crystals 106
4.2.2 SiC Layers 108
4.2.2.1 Monocrystalline Layers and Surfaces 108
4.2.2.2 Amorphous (a-SiC:H) and Polycrystalline Layers 110
4.2.3 SiC Fibers and Whiskers Ill
4.2.3.1 SiC Monofilaments Ill
4.2.3.2 SiC Whiskers Ill
4.2.4 SiC Nanopowders 112
5 SiC for High Temperature Applications 114
5.1 SiC Properties Related to Advanced Applications 114
5.2 Fabrication of SiC-Based Fibers and Structures 115
5.3 Properties of SiC Fibers and their Coatings 118
5.4 Properties and Applications of SiC Reinforced Glass
and Glass-Ceramics 120
5.5 Properties and Applications of C/SiC and SiC/SiC Composites . . 125
5.6 Porous SiC for Membranes and Filters 125
6 References 127
1
Introduction
Silicon carbide net-shaped ceramics are candidate materials for high-temper-

ature applications as engine and turbine parts and as cutting tools. They are also
important for the construction of reactors which show a high chemical
resistance. SiC material that reveals a well defined organization of its
microstructure is also a goal in the development of microelectronic devices.
The potential of silicon carbide pre-ceramic compounds was recognized
especially after Yajima et al. had prepared silicon carbide-based ceramic fibers
[1]. The development of tailor-made silicon carbide precursor molecules has to
be seen in close relation to the tremendously fast growth of the field of
Silicon Carbide - A Survey of Synthetic Approaches, Properties and Applications 61
organosilicon chemistry over the last few decades, which has provided powerful
tools for building up pre-ceramic networks. This progress enables the chemists
to design useful organosilicon polymers with a variety of molecular structures
that are built up from monomer units.
2
Synthesis of Precursor Molecules
2.1
Precursor-Concept
Making SiC-ceramics by pyrolysis of organosilicon polymers, or other

small organosilane and carbosilane molecules, has attracted considerable
attention.
SiC-precursor processing can be subdivided into gas (chemical vapor
deposition, CVD) and condensed phase methods. CVD, because of micro-
electronic applications, has spawned an entire field. It will be discussed here
only very briefly. We will focus more intensively on the use of silicon-
containing organometallic precursors as condensed phase sources of ceramic
materials.
The preparation of SiC fibers is an impressive demonstration of precursor
techniques. From this processing arises an important source of interest, but
the possibility of producing single phase and composite bulk bodies and thin
layers of SiC by polymer pyrolysis also gives rise to intensive research with
hopeful results to get the overall goal that enables us to delineate general
principles that guide choices of chemistries, processing methods, and material
properties, thereby permitting desired shapes with appropriate properties to
be produced. In this sense it seems to be necessary to review efforts in
chemical synthesis, polymer processing, and ceramic processing. The use of
organosilicon polymers offers a number of advantages at both the synthesis
and forming stages, some of which cannot be obtained in any other way
(fibers, coatings, membranes). The versatility of such organosilicon molecules
to form complex shapes is another characteristic that is valuable, particularly
for structural ceramic applications. This aspect demands the development of
Theologically useful precursors that can be shaped.
Very important is the opportunity to produce SiC material having a closely
controlled homogeneity (special optical and semiconducting properties).
Another important benefit of using organosilicon compounds is that
significantly lower processing temperatures can be employed (compared with
those for present inorganic raw materials). However, the use of lower
temperatures for SiC ceramic material formation has led to an unexpected
phenomenon: sometimes the incomplete partial pyrolysis of the precursor
provides a mixture of products instead of pure SiC.
Therefore the chemistry of the SiC precursor formation as well as that of the
pyrolysis plays a crucial role in the production of the "final" product.
A "useful" SiC fiber forming precursor must offer controllable rheology,
latent reactivity, controllable pyrolytic degradation, high ceramic yield, high
selectivity to SiC ceramic product, and controllable densification and

microstructural development, see [2].
Accessible precursor architectures (building units, dimensionality of the
polymer molecule) are dictated by the methods of the chemical synthesis of the
monomer units and the associated polymerization reactions. The type of
shape-forming procedure used (fiber-extrusion from solution or melt, spin-
coating, etc.) engenders constraints on what is considered useful polymer
rheology. Especially in the case of fiber-drawing or -extrusion the precursor
should exhibit thixotropic or non-Newtonian viscoelastic behavior. The
viscosity should be sufficiently high at zero shear such that once formed,
the material will retain its new shape.
To remain intact in shape during pyrolytic transformation the precursor
fiber or coating must retain some chemical reactivity so that the green bodies
can be rendered infusible before or during pyrolysis. Infusibility of
organosilicon precursor polymers is commonly obtained by extensive
cross-linking reactions via Si—X bonds (X = Cl, Br, H). In the case of low
latent polymer reactivity such a "curing process" can only be realized by
reaction with radical species which are generated for instance by photo-
assisted oxidation with oxygen. However, the resulting silicon carbide
involves silica impurities which diminish its quality. y-Ray as well as electron
beam irradiation can also be used to induce those desired molecular
rearrangements leading to infusible solids. Owing to their low selectivity
such reactions are difficult to control. The necessity to cure the green fibers
gives rise to a high cost level.
To aid processability or provide a higher latent reactivity level, the
precursor must commonly contain some additional substituents in the
polymer backbone. They give rise to eliminated gaseous products during
pyrolysis. The pyrolysis requires close monitoring to ensure conversion to
correct SiC material, to prevent retention of impurities or creation of
undesired pores.
SiC systems sometimes retain 1-2 wt% hydrogen at temperatures up to
1400 °C. This hydrogen content is responsible for the fact that the material will
not exhibit the properties of pure single-phase SiC. The hydrogen does not
permit normal microstructural development until it has been eliminated.
High ceramic yield is a critical criterion for the design and synthesis of
SiC precursors due to the excessive volume change associated with pyrolytic
conversion to SiC ceramic materials. Generally the precursor density is
much lower (~1 g cm"3) than that of SiC ceramics (2.5-3.27 g cm"3). That
means a ceramic yield of 50% is related to a volume decrease of 80%. So in
the case of low ceramic yield, achieving nearly net shape in the final ceramic
product becomes very difficult. However, this criterion is much less strong
in processing of films or fibers, since the shrinking is possible only in one
or two dimensions. Also gaseous by-products can leave easily from these
low dimension shapes, permitting ready densification at higher tempera-
tures.
Outgassing at higher temperatures can cause undesired cracking and
compositional changes, but it can also lead to the creation of small pores with
uniform size distribution. Such effects should be important in order to prepare

SiC membranes from thin precursor films.
Polymer films can be created from molten as well as dissolved organosilicon
compounds by dipping, spinning, or spraying on different substrates including
fused quartz or silicon.
If the precursor is targeted for coating applications only, where the
substrate provides most of the mechanical properties, the precursor must wet
the substrate effectively. Furthermore it seems generally necessary to limit
coating thicknesses to <1 um to get crackless layers.
Chemical and phase purity are also critical issues that drive SiC precursor
design. However, keep in mind that sometimes the same end product may
result from off-stoichiometry precursors or pyrolysis products by further
processing steps after precursor pyrolysis. Phase purity is quite important to
target such SiC materials which focus on reproducible semiconductor
qualities. However, the fact that H- and N-doped silicon carbide films behave
as high temperature semiconductors underlines that chemical and phase
purity are not always desirable.
Stoichiometric composition of the SiC-ceramics can be controlled by
variation of the Si/C-ratio in the precursor molecule.
2.2
Type of Precursors and Synthesis Principles
Scheme 1 presents important synthesis routes of silicon carbide precursor

molecules.
Within the circle characteristic polymer units are listed. Synthesis of tailor-
made polysilanes as well as polycarbosilanes provides a significant challenge to
chemists due to the difficulty in controlling the structure of such polymers. In
principle, the synthetic potential is not exhausted yet. Currently the metal
condensation routes are dominating. The most widely-used synthesis to date is
through the Wurtz reductive-coupling of dichloroorganosilanes; this is a
reaction with many attractive features. However, the successful recent search
of highly effective and selective catalysts should stimulate the perfection and
up-scaling of the alternative syntheses, especially dehydro-coupling and
hydrosilylation reactions.
2.3
SiC Synthesis from Small Precursor Molecules
Amorphous and microcrystalline films of silicon carbon alloy (a-SiC:H or

uc-SiC:H) are of considerable technical interest for their application as window
layers in hydrogenated amorphous silicon (a-Si:H) based solar cells [3], and
active layers in thin-film light emitting diodes [4]. They are usually grown by
plasma enhanced chemical vapor deposition (RF-CVD) with SiH4 + CH4 or
C2H2 gas mixtures in the "low power regime" [5]. The entire compositional range
of a-SiC:H films is obtainable. Nitrogen-doped amorphous hydrogenated
silicon-carbon alloy (a-SiC:N:H) films have been prepared by introducing
64 G. Roewer • U. Herzog • K. Trommer • E. Muller • S. Friihauf
X-Si—Si-X
II
disproportionation
H-Si-H
i
dehydrocoupling
hydrosilylation
-Si-H
I
chemical vapor deposition

-Si-C=c'
Scheme 1. Synthetic routes and building units of silicon carbide precursor molecules
nitrogen during the radiofrequency (RF) glow-discharge decomposition of

methane-silane-helium mixtures. Nitrogen doping enhances the photocon-
ductivity of the films, while the optical band-gap remains unchanged [6].
Another way of plasmachemical preparation involves the use of electron
cyclotron resonance for plasma assisted chemical vapor deposition (ER-CVD)
[7]. An alternative method to prepare undoped microcrystalline (fie) SiC:H
alloy films is developed recently as so-called hot wire chemical vapor
deposition (HWCVD) technique [8].
Using the plasma 13.56 MHz RF-CVD gas mixtures of SiH4 + CH4 and
SiH4 + C2H2 may be diluted with hydrogen. In this way the effect of hydrogen
radicals on properties and structure of hydrogenated silicon carbide films has
been investigated. Highly photoconductive a-Sii_xCx:H films can be prepared
by the hydrogen radical method.
Hot Wire Chemical Vapor Deposition (HWCVD-technique) yields device-
quality material of Si:C:H-alloy-films from pure methane and silane
mixtures.
Alternatively to decomposition of silane/hydrocarbon-mixtures hydroge-
nated amorphous Sii_xCx films can be deposited from the gas phase by
conventional pyrolysis, photolysis or plasma assisted decomposition of
various small hydridocarbosilanes. The composition and therefore the
properties of the films are known to be affected by the conditions of the
chemical vapor deposition and by the precursor structure. Regarding

the precursors, disilylmethane, trisilylmefhane [9], silacyclobutane [10],
1,3-disilacyclobutane [11], 1,3-disilabutane [12], 1,4-disilabutane [13],
tetramethylsilane [14], and monoorganylsilanes [15] have been tested.
Transversely excited atmospheric (TEA) CO2 laser-induced decomposition
of divinylsilane in the gas phase yields unsaturated C2-C4 hydrocarbons,
benzene and vinylsilane, and it represents a convenient process for chemical
vapor deposition of thin solid films composed of silicon carbide or
polycarbosilane [16].
In the case of using a-SiC:H as window material in p-i-n amorphous solar
cells this will not only reduce the interface reflections and surface absorptions
of incident light, but also greatly improve the transmissivity of the p layer in
the solar cells. By introducing a microcrystalline phase into a-SiC to obtain the
higher doping efficiency, the efficiency of the solar cells can be increased
further [17].
Structured (single crystalline, polycrystalline, or microcrystalline) semi-
conductor-grade SiC layers are produced primarily by chemical vapor
deposition [18], vapor-phase epitaxy (VPE) [19], molecular beam epitaxy
(MBE) [20], glow discharge [21], or sputtering processes [22].
Doping can be realized by adding compounds of boron or phosphorus [23].
The stoichiometric Si/C-ratio can be controlled within the limits a-Si:H to
a-C:H [24]. Hydrogen plays a dominant role to stabilize the amorphous
structure of the layers.
Routes that involve polymeric precursors are especially considered.
Poly(methylsilane), produced by the (j75-C5H5)2Zr(CH3)2-catalyzed dehydro-
coupling of CH3SiHCl2, can be pyrolyzed to silicon carbide and deposited as
thin films on silicon single-crystal wafers to form heterojunctions. Butler et al.
have developed procedures for the in situ doping of the growing SiC layers
from polymeric precursors, which can be performed in the same time as the
pyrolysis [25]. The intrinsic functional groups on the polymers (Si-H, Si-Cl,
Si-CH3) were utilized to produce SiC layers and in situ nitrogen-doped SiC
layers (n-type materials) coated on Si single-crystal wafers. The extent of
nitrogenation of the polymer can be controlled by varying the NH3 partial
pressure, and products with a wide range of resistivities were obtained.
Ternary Si:C:H films can also be gained by evaporating permefhyl-
polysilane powder which is stable even in air [26]. A growth rate of several
micrometers per minute can be obtained.
Until now there have been few attempts to prepare such a-SiC:H layers via
spin-coating or dip-coating of molten organosilicon polymers and a subse-
quent pyrolysis process [27]. Polycarbosilanes were dissolved in hexane, and
the resulting solutions were spin-coated onto silicon wafers, silica glass,
sapphire wafers, and graphite plates. These samples were heat treated under
vacuum (~10~ 4 Pa) at temperatures ranging from 200 to 1200 °C. In the range
1000-1200 °C, the initially amorphous films were transformed to polycrystal-
line /?-SiC [28]. Instead of employing conventional annealing at high
temperatures in an inert atmosphere, ion irradiation can also be used to
achieve the polymer-to-ceramic conversion [29].
66 G. Roewer • U. Herzog • K. Trommer • E. Miiller • S. Fruhauf
2.4
Polysilanes as Synthons Towards Silicon Carbide
Polysilanes have been the first class of precursors used to prepare silicon
carbide ceramics. In all cases, on pyrolysis, polysilanes are converted into
polycarbosilanes by a Kumada rearrangement prior to the formation of an
amorphous silicon carbide network. Several synthetic routes including
dehydro-polymerization, ring-opening polymerization of strained cyclosil-
anes, polymerization of masked disilenes, or base catalyzed disproportiona-
tion reactions of disilanes have been described to prepare linear or branched
polysilanes but despite its drawbacks the Wurtz-coupling route remains the
method applied most, especially to prepare linear polysilanes.
2.4.1
Polysilanes via Wurtz-Coupling
The Wurtz-coupling reaction of diorganodichlorosilanes is still the method

applied most to synthesize non-branched polysilanes [30].
R
I 2 n Na [toluene, 110 °C]
n Cl—Si—Cl
- 2 n NaCI
R R
The mechanism of this reaction is still discussed controversially [31]. A

problem is the polymodal molecular mass distribution. Usually three fractions
are obtained: cyclic oligomers with n = 5, 6, a fraction with intermediate chain
lengths (n = 30-50), and a high molecular fraction with Mw = 105-106 g mol" 1
[32]. A fractionation is achieved by a stepwise precipitation of the polysilanes
from a solution in toluene with an alcohol like methanol or isopropanol.
The yield of high molecular polysilanes is usually low (10-25%). Many
attempts have been made to increase the yield of the high molecular mass
polysilane by variation of the reaction conditions including inverse addition of
a sodium suspension to the dichlorosilane, reactions in other solvents such as
THF and at lower temperatures [33], addition of crown ethers [34] to increase
the solubility of the alkaline metal, and the use of ultrasound [35] to accelerate
the reaction.
The first reports about polysilanes made by Wurtz-coupling date from the
early 1920s when Kipping reacted diphenyldichlorosilane with sodium, but the
materials were not well characterized [36]. A quarter of a century later
Burkhard reported the synthesis of polydimethylsilane and the cyclic oligomer
(SiMe2)6 [37] by Wurtz-coupling of Me2SiCl2. Polydimethylsilane is described
as a white powder, extremely insoluble in organic solvents, which decomposes
without melting when heated above 250 °C. Further pyrolysis yields no
ceramic yield: the product disappears completely at 650 °C. Yajima et al.
described the preparation of a polycarbosilane [MeHSi-CH2]x initially from
dodecamethylcyclohexasilane [38] but later also from polydimethylsilane [39]
by thermal treatment either in an autoclave at 450-470 °C for 14 h or by

addition of python, a polyborosiloxane [40], and thermal treatment at 350 °C
for 10 h under N2 without an autoclave. The molecular weight distribution of
the polycarbosilane and the reactivity towards oxygen depend on the amount
of borodiphenylsiloxane (BDPSO) added. The larger the amount of BDPSO,
the higher the reaction temperature and the longer the reaction time, the larger
the molecular weight of the polycarbosilane becomes, and the higher its
stability towards oxidation [41].
The conversion of the polysilane into a polycarbosilane proceeds via a
Kumada rearrangement [42]:
Me Me Me H Me
I I
-Si — S i - Si — C — S i - (2)
I I
Me Me H H Me
It is assumed that this rearrangement proceeds via a radical cleavage of the

Si-Si chain and subsequent shift of a proton of a methyl group resulting in a
silaethene and finally addition of the Si—H bond at the Sr=C double bond
[43]:
CH, CH, CH3 CH3
II
-Si —Si — -Si« Si—
I I
CH, CH3
(3)
Me Me CH3 H CH3
I I
-Si —H Si- -Si- C— S i -
I // I
CH, H H
Me CH.
The as formed polycarbosilane with molecular weights in the range

Mn = 1000-2000 g mol" 1 and yields around 60% [44] should have the ideal
formula [MeHSi-CH2]n = "SiC2H6" but NMR investigations have shown the
presence of two different Si units: SiC3H (<5«-17 ppm) and SiC4 (<5«0 ppm)
with a molar ratio close to 1:2 [45], and typical elemental analysis yields
SiCi.77H3.7O0.03 suggesting a highly cross-linked polymer.
West et al. described the synthesis of "polysilastyrene", a copolymer of
dimethyldichlorosilane and phenylmethyldichlorosilane [46], the first polysi-
lane which is soluble in organic solvents:
Me Ph Me Ph
I 2(n+m) Na I I
n Cl — Si — Cl Si- •Si —
(4)
Cl — Si — Ci -2(n+m) NaCI
Me I Me Me
Me
Polysilastyrene forms silicon carbide directly on heating up to 800 °C

(10 K min" 1 ) with a ceramic yield of 30%, in contrast to polydimethylsilane
where a polycarbosilane intermediate has to be isolated first. Since then,
numerous polysilanes with a large variety of side chains have been prepared,
and their properties have been studied.
It has been shown, that the ceramic yield of polymethylphenylsilane (PMPS)
does not depend on the molar mass of the polymer, but can be increased either
by introduction of MeSiH or PhSiH units which provide crosslinks during
pyrolysis or by copolymerization with MeSiCl3 yielding a branched polymer.
The difference in the pyrolytic yields between poly(dimethylsilane) and phenyl
substituted polysilanes like polysilastyrene and polymethylphenylsilane
(PMPS) has been discussed as the result of two factors. First, the bulky
phenyl group may retard the depolymerization reaction, and second, the loss
of phenyl radicals followed by hydrogen abstraction and radical coupling
reaction to form polycarbosilanes [47].
West et al. have also tested the introduction of silacyclopentanes and allyl
side chains to get crosslinked polymers upon exposure to air or photochemical
crosslinking [48] (Schemes 2 and 3).
Wurtz coupling
or copolymers:
Si
homopolymer
crosslinking in air at 70 °C (2 h)
Scheme 2.
Wurtz coupling UV (300 nm)

crosslinked
Cl—Si—Cl polymer
Scheme 3.
Wurtz-coupling of methyldichlorosilane either with potassium in THF [49]

or with sodium [50] results in polymethylsilane (PMS). The polysilane is in
fact a branched polymer [(MeSiH)x(MeSi)i_Jn whose properties depend
largely on the value of x. For x values in the range of 0.65-0.85 the polymers
are soluble but the ceramic yields are quite low, 15-25%, and the ceramic
product contains excess silicon that crystallizes at higher temperatures. PMS
with x«0.40 gives ceramic yields around 60%, but the polymer is only
partially soluble.
Polysilynes, (RSi)n (R = Me, rc-Pr, w-Bu, n-Hex) have been reported by
Bianconi et al. [51]. They have been prepared by reduction of the trichloro-
silanes RSiCl3 using Na/K alloy and ultrasound. The kind of the alkyl
substituent has a major effect on the polymer properties. While the

poly(methylsilyne) is completely intractable, poly(hexylsilyne) is very soluble
in common organic solvents. No crystallinity could be detected by X-ray
diffraction, supporting a random network. It might be composed of 2D sheet-
like structures but it is more likely that polysilynes resemble dendrimers
having 3D rather than flat structures. Therefore, polysilynes represent
structural intermediates between polysilanes and elemental silicon [52].
A further increase of branching has been achieved by incorporation of
tetrafunctional monomers [53]:
Li [THF]
Si Si (5)
-LiX
X=CI, Br
Copolymerization of SiCl4 with Me2SiCl2 was less successful because this

reaction is difficult to control, due to favored formation of oxidation-sensitive
silicon clusters, and the reaction is often incomplete providing only low yields
of soluble materials consisting of a high proportion of low molecular
oligomers.
2.4.2
Disproportionation of Disilanes
The so-called disilane fraction is a by-product of the direct synthesis of

methylchlorosilanes from silicon and methyl chloride. This fraction is
obtained in 4-8% yield, which corresponds to more than 50,000 tons per
annum and consists mainly of 1,1,2,2-tetrachlorodimethyldisilane and
1,1,2-trichlorotrimethyldisilane. This mixture can be converted into branched
oligo- and polysilanes and methylchloromonosilanes by a Lewis-base catalyzed
disproportionation. Investigations by Trandell and Urry [54] on pure Si2Cl6,
Si2Cl5Me, and SiCl2Me-SiCl2Me have shown that disilanes with SiCl3 units are
disproportionated by NMe3 much faster than SiCl2Me-SiCl2Me and that
branched oligosilanes are formed:
ci ci CI SiCI3
NMe3, r.t.
4 CI —Si—Si —R 3 CI —Si—R CI3Si—Si—SiCI2R (6)
I
CI CI R = CI, Me CI SiCL
ci CI CI SiCI2Me
I NMe, I
Me-Si — Si —Me CI —Si—Me Me—Si —SiCI2Me (7)
17d/65°C
CI CI CI SiCLMe
Several catalysts were tested, and Baney et al. used 1 mass% Bu4PCl to
disproportionate a disilane fraction, and also pure SiCl2Me-SiCl2Me as well as
SiCl2Me-SiClMe2. The reactions, carried out under argon until 250 °C, yielded
in the first case 14.8% of a yellow glassy solid with the empirical formula
(MeSi)5.7(Me2Si)i 0Qi.9 [55]. Besides phosphonium and ammonium halides
HMPA, imidazol and pyrazol derivatives proved to be valuable catalysts for the
disproportionation reaction. N-Methylimidazol catalyzes the disproportiona-
tion of SiCl2Me-SiCl2Me even at room temperature, and also SiClMe2-SiClMe2
can be converted into Me2SiCl2 and a,co-dichloropermethyloligosilanes
(Si3—Si5) using this catalyst. The structure of the observed oligomers as well
as the nature of the monosilanes and the observed graduation of reactivities of
several methylchlorodisilanes suggest a mechanism via the formation of base
stabilized silylenes and subsequent insertion into Si—Cl bonds of other
disilane or already formed oligosilane molecules as is discussed in detail in
[56]:
Me
Cl | Cl
Cl Gl Cl Cl Cl Cl Si Cl
Me—Si—Si—Me
cat. I I I
Me—Si—Si—Si—Me
\
Cl—Si—Si—Si—Cl
I /
Cl Cl
- MeSiCI, I I
Cl Me Cl Me Me Me
Cl Cl Cl Cl Cl Cl (8)
\ / \ / \ /
Me—Si Cl Me-Si Si—Me
\ / \ /
Me-Si—Si—Me Me—Si—Si—Me
/ \ / \
Me—Si Si—Me Me—Si Si—Me
A A A A
Cl Cl Cl Cl Cl Cl Cl Cl
For further processing of the methylchloropolysilanes it is crucial to obtain
the product free from traces of the disproportionation catalyst. Otherwise the
disproportionation will proceed further, e.g., during melt spinning to get
polysilane fibers. Therefore several catalysts bearing HMPA-like structures or
nitrogen heterocycles have been fixed on a silica carrier [57] (Scheme 4).
Scheme 4.
The disilane fraction or pure SiCl2Me-SiCl2Me is heated to reflux and

condenses in the catalyst bed. A stream of argon removes the volatile
monosilanes formed, while the oligosilane mixture flows back into the flask.
An improvement of the rheologic properties of the polysilane has been
achieved by co-polymerization of SiCl2Me-SiCl2Me with styrene or by
introduction of phenyl substituents into the polysilane structure by dispro-

portionation of SiCl2Me-SiCl2Me in the presence of PhSiCL, [58]:
Cl Cl Cl Cl Cl Cl Cl Cl
oat.
1
Me-Si—Si—Me +
1
Ph—Si—Cl =»
1
Ph—Si— Si—Me
1 1
Ph—Si—Si—Si—Me
- MeSiCI3
C, c, Cl Cl C,
i Me Cl
Me (9)
Cl I/CI
Cl •"Si Cl
\
Cl—Si—-Si-Si-CI
1 /
1 1
1 Me Me
Ph
Besides the direct conversion of methylchlorodisilanes into polysilanes via

oligomeric intermediates there is also the possibility to substitute all chloro
functions of the methylchlorodisilanes by alkoxy groups [59]. Alkoxymeth-
yldisilanes can be disproportionated in the presence of strong bases like
lithium or sodium alkoxylates [60] or thermally resulting in essentially the
analogous structures of oligosilanes and monosilanes [61]:
Me Me Me
I I 220 °C I
MeO—Si—Si—OMe MeO -Si- -OMe (10)
I I - Me 2 Si(OMe) 2 I
Me Me Me
n = 3, 4
OMe
OMe
MeO OMe Me-Si OMe
200 °C /
Me-Si—Si—Me Me-Si—Si—Me (11)
- MeSi(OMe)3 | \
MeO OMe Me-Si OMe
OMe
OMe
OMe
MeO | ,OMe
MeO OMe MeO ^ S i OMe
I I 200°C \ | /
MeO—Si—Si—OMe MeO -Si—Si—Si—OM (12)
I I -Si(OMe) 4 / | \
MeO OMe MeO ^ S i OMe
MeO | OMe
OMe
These thermally induced reactions are thought to proceed via silylene

intermediates. Trapping products of silylenes, e.g., with diphenylacetylene,
could be observed. More recent publications describe the alkoxylate catalyzed
disproportionation reaction to proceed via silanide anions (MeO)2MeSi~ or
(MeO)Me2Si~ [62].
Kalchauer et al. obtained methoxymethylpolysilanes from stoichiometrically
different mixtures of tetramethoxydimethyldisilane and trimethoxytrimethyl-
disilane with NaOMe as catalyst at 160 °C. The properties of the resulting
polymers depend strongly on the molar ratio of the two disilanes used. While
an initial content of 70% Si2Me2(OMe)4 resulted in a polymer with Mw of
6600 g/mol (PD = 3.9) and a melting point of 200 °C, a disilane mixture
containing 90% Si2Me3(OMe)3 yielded a product with a Mw of 2900 g/mol
(PD = 2.2) and a melting point of 30 °C. Phenyl groups could be introduced
into the polymer by adding l,2-dimethyl-l,2-diphenyldichlorosilane to the
methoxymethyldisilane mixture [63].
2.4.3
Polysilanes via Dehydro-Coupling
Silanes bearing Si—H bonds are of basic interest in catalytic reactions.

Transition metals or transition metal complexes, respectively, are able to
catalyze hydrosilylation which is a useful approach to polysilanes. The major
drawback of this method is the relatively low molecular weight of the
polysilanes generated. The reaction pathway, as well as the reaction products,
depend on the nature of the educts and particularly on the catalyst which is
used. Numerous transition metals can be effective catalysts either as metal or
in compounds. For a general overview some key reactions of transition metal
catalysts are given in Scheme 5. The hydrosilylmetal species (a), which is
Si —X — Y — H
M1: group 8-10 metal, M2: group 3-4 metal, M3: group 10 metal
a: oxidative Addition, b: hydrosilylation, c: a-substituent migration, d: fS-hydride
elimination, e, f: reductive elimination, g: insertion of unsaturated compounds, h:
deydrocoupling
Scheme 5.
primarily formed via an oxidative addition can undergo a hydrosilylation with

unsaturated compounds (b) [64a-c], but an a-substituent migration (c)
[65a-d] or /^-hydride-elimination (d) [65a] are possible as well. Using group
10 metal complexes (M3) silicon-silicon bonds can be generated (f) [66a-d],
or the insertion of unsaturated compounds is possible (g) [67a-f]. These
reactions precede the formation of a disilylmetal species. To form Si—Si bonds
via dehydro-coupling (h) group 4 metals (M2) are the most effective catalysts.
The formation of Si—C bonds via hydrosilylation reactions (b, d, g) is
described in Sect. 2.5.3.
The catalysts which are effective for dehydro-coupling reactions are
classified in two broad groups by Gauvin et al. [68]. The first group are
cyclopentadienyl derivatives of the general formula Cp'xMRy (x = 1-2, y =
0-2). M is an early transition metal (Ti, Zr, V), lanthanide (La, Nd, Sm, Y, Lu)
[69], or actinide (U, Th) [70], R is usually methyl or hydride; other groups can
be benzyl [71], aryl [72], alkoxy, and silyl [73a,b] and Cp' stands for f75-C5H5,
775-C5H4Me or Cp* (^5-C5Me5).
Group 4 metal catalysts, primarily investigated by Harrod et al. [74a-d], are
the most important and best investigated class of catalysts for the formation of
silicon-silicon bonds via dehydro-coupling (Scheme 5h, Eq. 13):
R
cat
n H—Si — H Si—|—H + (n-1) H2 (13)
H H
Using this route, Harrod et al. prepared a poly(methylsilane) from

methylsilane, which gives after pyrolysis at 1000 °C in Ar atmosphere a
ceramic yield of 77% [75]. The exchange of the gaseous methylsilane against
the liquid 1,2-dimethyldisilane leads to a similar polymer which forms after
pyrolysis SiC in 88% ceramic yield [68]:
Me Me Me
Cp.TiMe, I 1000 °C
n H- -Si—H — -~ H- -Si- -H - SiC (14)
-m H, (Ar)
ceramic yield: 77 %
Me
I Me Me"
H — Si — H Cp2TiMe2 I 1500 °C
I -Si- -Si- H - SiC (15)
H — Si —H •mH, I (Ar)
I H
Me ceramic yield: 88 %
Via dehydro-coupling, a branching of Si-H containing polysilanes,

prepared by another method, is also possible. The work of Seyferth et al.
has shown that a polysilane precursor of low ceramic yield (about 20%) can
be transformed into an effective precursor with a high ceramic yield of
60-80% [76a, b]:
Me Me
I -Si-
Cp 2 TiMe 2
crosslinked polymer
1500°C
*- SiC
-Si-
- m H, (Ar)
(16)
H
ceramic yield: 72%
The most intense investigations were carried out in the development of

catalysts and in studies of the reaction mechanisms. To date there are at least
six mechanistic proposals for the reaction pathway of the dehydro-coupling,
which have been reviewed by Gauvin et al. [77]: two for catalysis by later
transition metal complexes, and four for catalysis by group 4 metallocenes.
Although the mechanism is still under discussion, the basic principle can be
depicted in Scheme 6 [78a-e].
RSiH.
M — HJ
M —H M — SiH,R
RhLSi-rH
RH2Si-SiH2R RSiH,
basic principle of c-bond metathesis
Scheme 6.
An unsaturated metal hydride derivative which is formed during an

induction period is postulated as the catalytically active species. The
mechanism includes two er-bond metathesis steps, each passing through a
four-centered transition state.
The primarily investigated group 4 metal catalysts show some general
trends considering the reaction products [79a, b]:
- Decreasing of the hydrosilane reactivity in the order of primary >
secondary > tertiary
- Generally atactic polymers
- Low and non-monomodal dispersity
- Linear polymers rather than branched ones
- High selectivity
Concerning the improvement of catalytic activity, especially to increase the
molecular weights and obtain efficiency for special problems, e.g., formation of
cyclic or linear products [79a, 80a, b], stereoselectivity [81], or addition of
secondary educts [80b], a large number of catalysts has been synthesized and
investigated. Because of a careful design, nowadays catalysts based on

cyclopentadienyl derivatives are able to control the polymer constitution (ring
or chain formation) as well as the molecular weights. With the 'cation-like'
early transition metal combination catalysts (e.g. Cp2MCl2/«-BuLi), developed
in the past few years, molecular weight distributions up to nearly 14,000 g mol" 1
were reached [79b]. They improve the yields of linear polymers and suppress
the depolymerization and formation of cyclic silanes. The obvious advantage is
the in situ generation of the catalytic species from metallocene dichloride and
H-butyllifhium or sodium bis(2-mefhoxyethoxy)aluminum hydride [82a-c].
The second class of catalysts can be described by the general formula LnM,
where M is a later transition metal (Cr, Mo, W, Rh) and L is carbonyl, olefin,
phosphine or halogen or combinations thereof [83a-c]. A famous example for
the LnM-group is Wilkinson's catalyst RhCl(PPh3)3, which catalyzes the
dehydro-coupling of PhMeSiH2 as well as Ph2MeSiH to give di- and trisilanes.
Other metals and ligands have also been tested but the results have mainly
been di- and trimers, which disqualifies this group of catalysts for the
preparation of preceramic polymers.
2.4.4
Other Synthetic Routes to Polysilanes
Besides the Wurtz-coupling reaction, other synthetic routes towards non-

branched polysilanes have been reported, which overcome the drawbacks of
polymodal molar weight distribution and low yield of the high molar mass
fraction. However, so far the application of these routes has been very limited.
The anionic ring opening polymerization of the strained 1,2,3,4-tetramethyl-
1,2,3,4-tetraphenylcyclotetrasilane leads to poly(methylphenylsilane) in high
yield but the starting cydotetrasilane has to be prepared in three steps from
diphenyldichlorosilane via octaphenylcyclotetrasilane [84]:
Ph Ph TfO Ph Me Ph
I I II II
Ph-Si—Si—Ph 4HOTf Ph-Si—Si—OTf 4 MeMgBr Ph-Si—Si—Me
I I I I - I I (17)
Ph-Si—Si—Ph -4PhH TfO—Si—Si—Ph -4BrMgOTf Me-Si—Si—Ph
II II II
Ph Ph Ph OTf Ph Me
Silicon-silicon bonds can also be formed by electrochemical reduction of
chlorosilanes [85], a method which has been applied first to form disilanes
from triorganochlorosilanes [86].
Poly(dimethylsilane) can be synthesized by electrochemical reduction of
dimethyldichlorosilane in 30% yield [87] and poly(methylphenylsilane) has
also been synthesized by reduction of PhMeSiCl2 [88]. In this method the
problem is the oxidation of the anode by elemental chlorine to form metal
chlorides, or the formation of corrosive hydrogen chloride [89]. Therefore the
electrochemical polymerization of hydrosilanes like phenylmethylsilane,
phenylsilane, or hexylsilane has also been investigated. Relatively low
molecular weight polymers could be obtained in 32-70% yield [90]:
electrolysis
(18)
0.2 M Bu4NBF4/DME
2 F/mol
Sakurai et al. [91] have developed a synthetic route towards polysilanes

via masked disilenes (7,8-disilabicyclo[2.2.2]octa-2,5-dienes). The starting
materials are accessible from dichlorodisilanes and lithium biphenylide [92]
(R1, R2, R3, R4 = Me, Pr, Bu, Hex):
CISiR 1 R 2 -SiR 3 R 4 CI
(19)
2 Ph
- PhPh - 2 LiCI
Alkyllithiums such as n- or sec-butyllithium or phenyllithium serve as

initiators:
R1 R3
RLi [THF] -70 °C
-Si — Si- (20)
- n PhPh
R2 R4
Reported polymer yields of 60-80% are much higher than by the Wurtz-
coupling synthesis, and the resulting polymers have Mn values in the range of
104-105 depending on the amount of RLi used to initiate the polymerization.
Furthermore GPC profiles reveal a monomodal distribution of molecular
weights and polydispersities between 1.2 and 1.9.
2.5
Polycarbosilanes
As is obvious from the case of polymethylsilane, the use of a polymer that has
a Si/C ratio of 1:1, but where the carbon is located in the side groups, does not
necessarily lead to a ceramic product which retains the starting Si/C ratio. A
potential solution to this loss of carbon on pyrolysis is to incorporate it as
bridging methylene groups in the polymeric structure where it is effectively

locked between two silicon atoms. Whereas the term "polycarbosilanes" in its
wider sense incorporates all polymers which contain both carbon and silicon
atoms in its main chains, a true polycarbosilane possesses a chain of
alternating methylene and silylene units. Three general approaches to
polycarbosilanes have emerged, the direct polymerization of bifunctional
monomers via Grignard coupling, transition metal catalyzed hydrosilylation
reactions, and the ring opening polymerization of pre-formed strained sila-
and 1,3-disilacyclobutanes.
2.5.1
Polycarbosilanes via Grignard Coupling
Polycarbosilanes can be obtained by a Wurtz-like condensation of dichloro-

silanes in mixture with a dihalomethane. The products obtained in these
reactions do contain not only the desired Si-C-Si linkages but also oligosilane
units -(SiR2)x- and -(CH2)X- units [93]:
R R H
4Na I I
Cl — Si — Cl + CH2Br2 -Si — C (21)
- 2 NaBr
R - 2 NaCI H
In the case of R = Ph further derivatizations have been performed [94]

yielding polycarbosilanes with unsaturated side chains [95] as well as a
polysilaethene (Scheme 7). The ceramic yield of the latter could be increased
up to 91% if some of the phenyl groups remained [96]. Branched polycar-
bosilanes were obtained by a Wurtz condensation of mixtures of PhSiCl3 and
Ph2SiCl2 with CH2Br2 [97].
Using similar procedures polycarbosilanes could be obtained which were
transformed into nearly stoichiometric silicon carbide [98]. In order to obtain
higher ceramic yields the precursors were thermolyzed at 300 °C for 3 h under
argon prior to pyrolysis. Crosslinking of the polymer chains occurred during
the thermolysis under elimination of hydrogen and methylsilanes (Scheme 8).
Besides polycarbosilanes possessing CH2 units, precursors with the
unsaturated —C=C— and —C=C—C=C— linkages have attracted some
attention because the reactive triple bonds facilitate further crosslinking to
prevent a remelting during pyrolysis (Scheme 9). The products give high
ceramic yields, but due to the high carbon content of the precursors, the
pyrolysis products contain much excess carbon. The utilized lithium acetylide
can be made either from BuLi and C2H2 or better from CC12=CHC1. The latter
procedure has the advantage that no Li—C=C—H is formed which would
result in terminal units and hence in low molecular mass of the polymers [99].
Analogously Li2C4 is accessible from perchlorobutadiene and BuLi [100] and
polycarbosilanes with —R2Si—C=C—C=C— units have been reported [101].
Freshly prepared polymers are colorless or light yellow but on standing they
turn red or brown. This is accompanied by crosslinking reactions and a
decrease in solubility.
Ph H Cl H
I I HCI 1 1
Si—C
-Si—C- [Aicy 1 1
I I Cl H
Ph H
LiAIH, RMgX Cl2 (UV)

1
H H R H Cl Cl
I I 1 1
—Si—C- — Si—C 1 Si—C—
I 1 1
H H R H Cl Cl
n
Mn = 690 R = Vi, All, ...
R'Li
PD = 9.4 Mn = 800... 1850
ceram. yield = 40.3 % PD = 30 - 60 R1 R1
ceram. yield = 50-60 % I I
-Si—C—
R1 = Et, Pr, Bu, Pent, Hex

Mn = 830 (Et)... 1660 (Hex)
PD= 1.6-2.4
Scheme 7.
Besides chlorosilanes, silyl triflates have also been used as starting

materials offering an approach to highly ordered polycarbosilanes [102]
(Scheme 10).
Also phenylene units have been introduced into polycarbosilane chains
either by polymerization of a para-bromo phenylsilane [103]:
n
Mg
Br—(Cjy~ Si—OEt Si — (22)
I
OEt OEt
or by applying the silyl triflate chemistry:
R
I
BrMg -MgBr + TfO—Si—OTf (23)
- 2 MgBrOTf
Me Me
Suitably substituted carbosilanes allow the introduction of reactive vinyl

and Si-H side groups which can be used for transition metal catalyzed
crosslinking [104] (Scheme 11).
Crosslinking has also been achieved by introduction of strained silacyc-
lobutane rings into the polymer [105]:
"Pf
BiMg—(^j)— MgBr + ^ crosslinked polymer
(24)
X
Cl CI
M n =1430 PD=1.85 ceramic yield: 60 % (800-C)
Me Me H Me
2 Mg [Zn] I I I
Cl-Si —C—Si —Cl
I (TWF) I I I
H
H H H
Na (toluene) + PhMeSiCI2
110 °C Na (toluene)
110 °C
Me H Me Ph Me H Me "
-Si — C — Si-
I I I I
Si- ol
-Si-
I I I I I I
H H H Me H H H
0.4 -
Mn = 1400 PD = 3.1 HCI [AICI3]

ceramic yield = 14 %
LiAIH4
thermolysis
3 h at 300-350 °C
H Me H Me
I I I I
Mn = 2550 - 2750 PD = 3-3.3 -Si- — Si — C—Si-
ceramic yield = 60-80 % Me
I I I
H H H
Si: 47.5 mol-%
ceramic yield = 14.5 %
C: 51.3 mol-% at 1000 °C
O: 1.3 mol-% after thermolysis (3h at 250-300 °C):
ceramic yield: 64 %
Si: 45.5 mol-%
C: 51.5 mol-% \- at 1000 °C
O: 3 mol-%
Scheme 8.
2.5.2
Polycarbosilanes via ROP of Sila- and Disilacyclobutanes
Sila- and 1,3-disila cyclobutanes possess an increased reactivity because of

their ring strain [106] and can be transformed into polycarbosilanes by ring
opening polymerization using either thermal activation [107], anionic
polymerization with organolithium compounds [108], or mostly platinum
[109] or other late transition metal catalysts [110].
A detailed investigation of the reaction of several silacyclobutanes with
platinum catalysts has revealed that these reactions proceed via the
Cl Cl 3 BuLi
Cl -78 °C
2 Me.SiHCI CIMePSiSiMePCI Me,SiCI, PhMeSiCL Ph,SiCI,
Me Me Me Me Me Ph Ph
I i I I
H-Si- -Si-H Si~Si — -Si- -Si- -^—Si —
i i i i I
Me Me Me Me Me Me Ph
Mw = 70000 Mw = 20000 Mw = 20000 Mw = 30000

CCI.
PD = 2.6 PD= 1.7 PD = 2.2 PD = 2.9
Me Me ceramic yields:
Cl-Si- -Si-CI 58.9 % 80.3 % 81.4% 76.8 %
i I
Me Me (70.9 % SiC) (59.6% SiC) (35.5 % SiC) (29 % SiC)
Na
(xylene, 35 °C)
Me Me
I !
-Si- -Si —
Me Me
Scheme 9.
Me Me Me
2 CIMgCCH I 2 BuLi I
TfO—Si—OTf
H-C=C-Si—C=C-H l_i-CSC-Si-C=C-Li
Me
MePhSi(OTf)2
—CBC-Si—CSC—Si— —C=C-Si—CSC—Si—
- LiOTf
O= S =
Scheme 10.
oxidative addition of a Si—C bond at platinum under formation of a

l-platina-2-silacyclopentane. Insertion of a second silacyclobutane results in
a l-platina-2,6-disilacyclononane which can either eliminate the dimer
1,5-disilacyclobutane or produce poly(trimethylenesilylene) depending on
the nature of the ligands at platinum [111] (Scheme 12).
Molecular weights as high as M n = 94,000 g m o P 1 (PD = 1.8) have been
achieved with 0.002 mol.% Pt(cod) 2 within 2 h at room temperature and 90%
R H
Me Me
Me Me
-CH-
(f))—Si —Si—
Me Me
Scheme 11.
Pt°L 3 (CH2)3SiMe2
Si—Me
-L
Me
Me Me
- -PtL,- L = PEt3, PPh3 n (CH2)3SiMe2 L = olefin
Me
Si'
I
Me
Scheme 12.
conversion of the monomer. H2PtCl6 can also be used if higher reaction

temperatures (40 °C) are applied in order to reduce PtIV to active Pt° species
[111].
The anionic polymerization of 1,1-dimethyl and 1,1-diethylsilacyclobutane

with butyl lithium in THF/hexane (1:1) at -48 °C proceeds with a living
nature. M n increases in proportion as the yield of polymer increases. Addition
of new monomer to the reaction restarts the polymerization at the same rate as
in the initial stage.
The polymerization at —48 °C yielded polycarbosilanes in 99% yield with
Mn of 2400 g moP 1 and a low polydispersity of 1.1. Polymerization at 0 °C
yielded a higher Mn of 4400 g moP 1 but at the expense of the polydispersity,
which increases to 1.47.
The reaction is thought to proceed via hypercoordinated silicate ions [112]:
Me Me
M6 Mi
I O I Me U ' MC^
Me-Si—| BuLi ^Si—1 (CH 2 ) 3 SiMe z , ^Si—i (25)
| I - BU | | ir Bu-si-(CH2)3 I I i_r
Me
The formation of a pentacoordinated silicon atom reduces the ring strain of

the four membered ring [106].
Addition of styrene to the living polycarbosilane provided a poly(l,
1-dimethylsilabutene-co-styrene) block copolymer.
In contrast, a polymerization of 1-methyl-1-phenylsilacyclobutane in THF
at -78 °C did not show a living nature [112].
In order to obtain polycarbosilanes with a Si/C ratio of 1:1 in the main chain
1,3-disilacyclobutanes have to be used as starting materials. An early approach
to disilacyclobutanes utilizes the pyrolytic conversion of silacyclobutanes into
disilacyclobutanes under elimination of ethene [113]. The electrochemical
reduction of chloromethyldimefhylchlorosilane yields 1,1,3,3-tetramethyl-l,
3-disilacyclobutane in 34% yield besides other products [114]. A more recent
synthetic pathway starts from chloromethyltrichlorosilane. While a direct
Grignard coupling in analogy to syntheses of disilacyclobutanes by Kriner
[115] does not give the desired l,l,3,3-tetrachloro-l,3-disilacyclobutane in
satisfactory yields, the introduction of ethoxy groups to mask two of the three
reactive functions at silicon has overcome this problem [116]. The ethoxy
functions can be resubstituted for chlorine with acetyl chloride. The total yield
of l,l,3,3-tetrachloro-l,3-disilacyclobutane (TCDSCB) from ClCH2SiCl3 is
about 30% via this procedure. The resulting TCDSCB can be polymerized
with H2PtCl6 in benzene at 90 °C and hydrogenated with LiAlH4 in Et2O to
yield polysilaethene (PSE) with M n «10,000 g moP 1 determined by end
groups in the H-NMR spectrum and Mn ss 20,000 g moP 1 determined by GPC.
Alternatively l,l,3,3-tetraethoxy-l,3-disilacyclobutane itself can be poly-
merized followed by LiAlH4 reduction (Scheme 13). The resulting product has
a lower molecular weight of M w «7000 g moP 1 .
PSE has an extraordinarily high ceramic yield of 86-90% by TGA
(theoretical yield: 91%) and retains its initial Si/C ratio of 1:1 on pyrolysis
to give a nanocrystalline fS-SiC at 1000 °C.
Therefore PSE might be a useful model system for the study of the
precursor-to-ceramic conversion, despite its relatively high costs.
ci CI OEt
I CI, | 2 EtOH I
CI-CH,—Si-CI C I - C R , —Si - C I
CH3 — Si-CI -HCI - 2 HCI
CI OEt
CI
Mg [THF]
- MgCI2
CI CI OEt
I
uI
CI—Si 840 °C AcCI [FeCy EtO—Si —
CI—Si
30% 1
-Si - O E t
Si - C I I
CI OEt
[H2PtClj
90 °C [(olefin)PtCI2]
CI OEt
I LiAIH4 LiAIH4
-Si -CH, - Si - C H , -
I [Et2O] [Et2O]
CI OEt
Scheme 13.
A highly branched "(SiH 2 CH 2 ) n " is accessible by a Grignard coupling of

ClCH2SiCl3 followed by LiAlH4 reduction [117]. The product can be described
with a formula (H3SiCH2)w(H2SiCH2)x(HSiCH2)y(SiCH2)z. The amount of
the different units was evaluated by 29Si NMR spectroscopy to be
w:x:y:z « 11:20:8:2. Determination of the molecular weight by GPC yielded
Mw w 1000 g m o F 1 and M n w 600 g mol" 1 , corresponding to an average
degree of polymerization of about 13. This SiC precursor has proven useful as
SiC matrix source for SiC fiber and particulate-reinforced SiC/SiC composites.
Its production has been scaled-up to 6 I/batch [118].
Starting from l,3-dichloro-l,3-dimethyl-l,3-disilacyclobutane the ring
opening polymerization yields a polychloromethylsilaethene with high mo-
lecular weight which is readily soluble in a wide range of organic solvents.
Substitution reactions have led to a variety of differently substituted
polysilaethenes [119] (Scheme 14).
While the hydrolytic stability of the alkoxy substituted polycarbosilanes
depends very much on the nature of the alkoxy group (stability raises in the
order Ac < Et < CH2CF3 < Ph) the fluoro substituted one is stable towards
water as well as atmospheric exposure for extended periods. Whereas
(SiMeFCH2) is soluble in common organic solvents, the corresponding
difluoro substituted polymer (SiF2CH2)n is completely insoluble in all solvents
tested so far. This polymer is obtained either by ROP of cyclo-(SiF2CH2)2 or -
with a higher molar weight - by F-for-OR replacement from poly(diethoxy-
silaethene) with Et2O/BF3 [120].
LiAIH4
RONa
R = Et, CH2CF3, Ph, Ac
OR H
I
Si — Si —CH 2 —
I
Me Me
n
(R=Et)
CH2=CHR [Pt]
Et2O / BF3
CH2CH2R
-Si-CH2 — -Si — CH2 —
Me Me
Scheme 14.
2.5.3
Polycarbosilanes by Hydrosilylation
The hydrosilylation of silanes with unsaturated compounds (Scheme 5, b) is a

convenient method for preparing polycarbosilanes. The olefinic functions can
be delivered either from organic molecules or from olefinic substituted
polysilanes themselves. Corriu et al. explored platinum catalyzed hydrosily-
lation of vinyl and allylsilanes [121a-e]. NMR spectroscopic data revealed the
predominantly formed linear constitution of the resulting polymers (polyvinyl
compounds up to 85%, allyl compounds quantitatively). The polymerization of
ViHSiCl2 was optimized to give polymers with average molecular masses of
up to 5500 g moP 1 in yields of up to 78%. By treatment with LiAlH4 the
hydrosilylation products based on vinylsilanes were converted into
poly(silethylene)s (Eq. 26). Ceramic yields, depending strongly on the number
of vinyl endcaps, varied from 12 to 62%. Experiments starting from

dihydrogeno- and divinylsilanes result in similar products [121d]. To prepare
high molecular weight polymers, the hydrosilylation of ethynylhydrosilanes
is more suitable. Reactions like the one in Eq. (27) give polymers with
average molecular weights between 30,000 and 100,000 g mol" 1 [122]. The
conventionally used catalyst is H2PtCl6:
LiAlH,
(26)
Cl
R R
[H 2 Ptcy I 1200 °C
n H-C=C—Si—H -Si' SiC (27)
I I
R R
Using bifunctional organic molecules (e.g., 1,3-butadiene) and hydrosilanes,

the silicon bridging units in the polymer can be varied (Scheme 15). Because
of their low ceramic yield and the low thermal stability, polycarbosilanes
containing longer hydrocarbon bridges are not suitable for preceramic
polymers [123a, b].
Scheme 15.
3
Pyrolysis
3.1
Conversion of Organosilicon Precursor to Inorganic Material
The step by step thermal treatment of the polymer SiC-precursor in the

temperature range 200-1400 °C generally gives rise to a lot of chemical
200 °C Precursor: polysilane/carbosilane
400 °c cross-linking: condensation, introduction of heteroatoms

thermal rearrangement of skeleton
loss of volatilities
•Ei=-Si-'Si^p 2-3-dimensional network

~Si-E^Si- f
;-SisTE-si~
h
' ^E E= Si, C, heteroatom:N, O
500 °C ceramization
loss of hydrocarbons (CH4, C2H4...) and hydrogen
decomposition of side chains
900 °C amorphous cowalent ceramic
'Vc'- ?v ' o s s ° ' hydrogen

-sik.c_4i-
,C'^'^Q crystallization
1200 °C crystalline ceramic 3-dimensional

Scheme 16. Typical stages and products during the pyrolysis process of polysilanes and
polycarbosilanes
intermediates starting with further cross linking and ending up with crystalline
ceramic. Scheme 16 represents an overview over the characteristic stages and
their sequences depending on the pyrolysis temperature.
The processes occurring during cross linking and ceramization as well as
the structure of the preceramic network and the amorphous covalent ceramic
depend strongly on the chemical structure of the synthesized precursor
molecule. The influence of the type of the polymer molecule, of the catalysts,
steric effects of the side groups, bond strengths, as well as radical stabilities are
crucial factors.
Apart from the specific properties of the precursors the parameters used for
the ceramization steps like heating rate, atmosphere, and duration of pyrolysis
also influence the ceramic yield and composition as well as the evolution of
gaseous species during pyrolysis.
3.1.1
Polysilanes/Carbosilanes
The presence of functional groups on the polymer material is vital for the
cross-linking and doping steps, as well as for promoting adhesions to substrate.
Intrinsic functional groups on polysilanes (Si—H, Si—Cl,...Si—CH=CHR) or
additional built in functionalities must be utilized to "cure" the precursor, i.e.,
to stabilize the network and also to gain an increase of the ceramic yield. Small
amounts of reactive curing gases (O2, H2O, NH3, etc.) and/or a low level of
unsaturated side groups should give a strong cross-linking effect. From such a
curing process heteroatom doping also arises in the SiC-ceramic.
The thermal treatment of vinylic polysilane SiC precursor polymers involves
<300 °C cross-linking and loss of low molecular weight oligomers. Polymer
degradation, which is characterized by significant decomposition of the side
groups and chain scission, along with conversion of the polysilane skeleton to
poly(carbosilane), takes place between 300 and 750 °C [124].
The decomposition chemistry of most polysilanes above 300 °C is quite
complex and involves restructuring of the polymer backbone as well as
substantial cleavage of the side groups. Generally a poly(alkyl)silane must go
through a carbosilane intermediate prior to formation of the SiC network
(Kumada rearrangement) [125]:
volatile oligomers
C—H yS
^ — S'-H (28)
C
Si—Si —
The intensities of the Si-H stretching band (~2080 cm x) and the Si-CH2-Si
deformation band (~1050 cm" 1 ) of the IR spectrum increase with an increase
in pyrolysis temperature from 250 to 400 °C. Methylene insertion reactions are
also suggested by changes in the XH CRAMPS, 13C MAS, and 29Si MAS NMR
spectra. The observation of more than one distinct methylene carbon/proton
resonance is consistent with the complicated chemistry of the system, wherein
different reactions generate chemically distinct methylene environments.
High yields of SiC can be expected when cross-link formation is encouraged
so as to minimize loss of silane fragments by competing chain-scission
processes:
\ + HC"? — ? i (29)
-Si- "Si" "Si- ?
The presence of catalytically active amounts of (j75-C5H5)2Zr(CH3)2 is

particularly effective in favoring formation of Si-C rather than Si—Si bonds
during pyrolysis [25].
The SiC yield under argon atmosphere is very dependent on pendant groups
in both homopolymers and copolymers [126]. The yields were found to
increase as the poly(dimethylsilane) structure was substituted with hexyl and
especially phenyl groups. A significant improvement in SiC yield has come
from co-polymerization of low levels of silane monomers containing unsat-
urated groups (vinyl or allyl). These effects should be engendered mainly by
radicals, which are created as the temperature rises. As indicated by FTIR
spectra between 300 and 450 °C, a rapid loss of -CH3, -C6H5 and hexyl groups
takes place. At temperatures above 600 °C the remaining Si-CH3 groups are
lost, and there is a progressive conversion from the imperfect to perfect SiC
network. The escaping gases detected by TG/MS analysis arise from
condensation reactions between the various functional groups (Si—H, Si—CH3,
C—H, Si—CH2—Si).
ESR experiments also indicate that a substantial concentration of free
radicals is present in the solids, which corroborates the participation of radical
species during the decomposition. A variety of radical combination and
abstraction reactions are possible at high temperature and can account for the
various hydrocarbons and silane products observed. In general, hydrogen
atom as well as carbon or silicon atom centered radicals could be created (c.f.
Scheme 17).
1
-Si-H
i
-Si- 4- -H
1
1
Qi f"*M Qi - -Si- +• CH 2 -Si
Sl CH 2 Sl 1
-Si-CH 3 -Si- 4• CH3

1
Scheme 17.
In an early pyrolysis state (low temperature region) radical species should

represent relatively large units which are predominantly 2-3 dimensional, thus
incorporating cycles, adamantane-, or diadamantane-entities (Fig. 1).
Formation of such skeletons seems to be favored energetically as shown by
methylsilanes gas-phase pyrolysis results of Fritz et al. [127]. Some radicals
formed may abstract the hydrogen atoms to yield H2, CH4. The combination of
various silyl- and alkyl-radicals primarily formed leads to the formation of
Si—Si, Si—C, and C—C bonds, according to Scheme 18.
The driving force arises from the tendency to build up progressively a great
number of energetically favored units, which eventually develop into networks.
The site of actual bond formation (Si—Si, or Si—C, or C—C) should depend on
the steric arrangements which, in special cases, enables the realizations of
favored skeleton geometries. Furthermore the production of small molecules
^Si"/ -Si ^Si"/ ' -E

II I I
Fig. 1. Favored radical species related to the thermal decomposition of polysilanes or
polycarbosilanes (E = Si, C)
1 ! I i
R-C- + C-R -— R-C-C-R
1 1 I I
1 1 I I
R-Si- + •C-R —~ R-Si-C-R
1 I I i
I I I
R-Si- + •Si-R — R-Si-Si-R
I I I i
Scheme 18.
like hydrocarbons, various methylsilanes, oligosilanes, oligosilylmethylenes

via hydrogen abstraction of small C, as well Si atom centered radicals from
C—H or Si—H bonds must be considered [128], of course (c.f. Scheme 19).
i i i i
R-E- + H-Si- •-«- R-E-H + Si-
i I I I
I I I I
R-E- + H-C- - ~ - R-E-H + C-
I I I I
E=C, Si
Scheme 19.
Such easily escaping molecules were detected by TG/MS analysis.

Indeed, a significant initial cross-linking of the polymer prevents a high
silicon loss via fragments. The presence of Si—H bonds improves the
condensation to the ceramic product. Poly(carbosilane) examples resulting
from pyrolysis of monosilanes SiMexH4_x were given by Fritz et al. [127]. This
conclusion is also in agreement with results of Interrante and coworkers [129]
from investigation of polymers [-H2Si-CH2-]x.
Mineralization starts above 500 °C, with a large evolution of hydrocarbons,
e.g., methane in the case of methyl groups containing polymers. *H CRAMPS
spectra of the partially pyrolyzed polymers confirm the decreasing concen-
tration of hydrogen with increasing pyrolysis temperature. The 13C MAS
spectra demonstrate the coalescence of various SixCy units to a single peak
representing SiC4 units in SiC (and incorporation of excess carbon into the
resulting ceramic product). Hydrogen escaping at higher temperature prob-
ably results from C—H bond homolysis. The loss of hydrogen must give rise to
the formation of unsaturated carbon atoms and, consequently, to changed type
of the C-radicals (sp 3 —» sp 2 hybridized carbon atom).
The 13C chemical shift is a good indicator of the change in the carbon
environment during the conversion from polymer to silicon carbide. When the
pyrolysis temperature increases, there are fewer protons bonded to carbon,
and therefore the 13C chemical shifts move downfield [130] (formation of C=C
double bonds). By this effect the conversion of aliphatic into aromatic carbon
during pyrolysis is clearly indicated in the case of carbon-rich SiC. XH -> 13C
cross polarization NMR experiments indicate that, even after heating to
90 G. Roewer • U. Herzog • K. Trommer • E. Muller • S. Fruhauf
1000 °C, there remains a significant hydrogen concentration in the ceramic,

which may increase up to bonding saturation around C and Si centers.
Aromatic carbon seems to be one of the major structural defects in
amorphous SiC phases resulting from polymers. It stays in the SiC as
amorphous carbon up to 1300 °C as shown by Raman studies. Graphitization
of carbon usually occurs at temperatures higher than 2000 °C.
The concentration of dangling bonds can be analyzed by ESR. A g-value of
2.003 has been assigned to the C-dangling bonds. These bonds are a major
defect in the amorphous and microcrystalline silicon carbide, especially when
an excess of carbon is present. In general, the spin density of the pyrolyzed
samples increases with increasing firing temperatures, but reaches a maximum
at some temperature level, and drops significantly above this temperature. The
decrease of the spin density at higher temperatures is very likely due to the
crystallization of SiC.
When the organosilicon polymers are heated under hydrogen, both C=C
bonds and C-dangling bonds decrease significantly. This may be due to the
fact that H2 becomes H radicals at high temperatures.
However, the hydrogen has little effect on the structure of microcrystalline
SiC (presumably the a-SiC has to eliminate most of the hydrogen before
crystallization can really occur at >1000 °C).
3.1.2
Pyrolysis under Ammonia
In the case of poly(carbosilane)s, carbon removal and amination occur

simultaneously in the range 400-700 °C. The escape of hydrogen starts at
lower temperature than under argon. The first step of the amination process
corresponds to the nucleophilic substitution of hydrogen atoms bonded to
silicon by ammonia (Eq. 30):
-Si-H + NH3 —- -Si-NH 2 + H2 (30)
The escape of hydrogen carbons and organosilicon molecules starts at the

same temperature as under argon, which indicates that homolytic cleavages
take place. Accordingly, above 500 °C a radical amination mechanism should
be considered. An easy complete amination is observed. This may be explained
by hydrogen abstraction from N—H bonds, either on NH3 molecules or on
SiH2 and SiNHSi groups (c.f. Scheme 20).
R- + NH3 — R-H •t- NH2
Ft- + -Si-NH 2 — R-H 4- -Si-NH
. 1
R- + -Si-NH-Si- — R-H H- -Si-N-Si
i I I I
Scheme 20.
Thus, in the case of pyrolysis under ammonia an abundance of H atoms

arises from the N—H bonds. Therefore reactions depicted in Scheme 20
compete efficiently with other H-abstraction reactions (Scheme 19). These
reactions lead to the consumption of the radicals, which hinders the cross-
linking of silicon atoms via Si-Si or Si-C bridges, thus avoiding the formation
of amorphous silicon carbide and allowing the complete amination of the
polymer. Sometimes the loss of organosilicon species is noticeably lower under
ammonia than under argon. This behavior may be explained by the early
amination of Si—H bonds to give Si-NH2 groups which then condense through
transamination reactions leading to a cross-linking of the chains by Si-N-Si
bridges (200-500 °C).
3.1.3
Poly(chloromethylsilanes/carbosilanes)
Poly(methylchlorosilanes) which have been prepared by Lewis-base catalyzed

disproportionation of 1,1,2,2-tetrachlorodimethyldisilane give rise to a
complex pyrolytic conversion into silicon carbide. In the temperature region
180-450 °C under dry argon, the polymer samples are cross-linked. Charac-
terization of the pyrolysis intermediates and final products by 13C and 29Si
MAS in both CP and IRCP techniques as well as TG/MS studies strongly
suggest the loss of methylchlorosilanes as well as oligosilanes during pyrolysis
(c.f. Scheme 21).
Me „. Cl Me Me Cl ,.
\ C \ / \ / Me
\Si S i/ ^ _ , n \ S i S Ii ^ \ I I
I \ Cl / \ -MeSiCI3 1
Me Me Me
Scheme 21.
Evolution of HC1 at low temperature levels beside of CH4 and H2 is observed

(c.f. Scheme 22).
Results of these investigations allowed one to propose two different
mechanisms for the generation of carbosilane units in poly(chloromethyl)
silanes.
i i ii
-Si-H + H-Si- -Si — Si-
i I II
-Si-CI + H-Si- -Si —Si- + HCI
-Si-CH3 + H-Si- -Si —Si- + CH,

I I II *
Scheme 22.
At first - at low temperatures (T > 180 °C) carbosilane moieties

Si—CH2—Si arise via condensation steps (Eq. 31):
-Si-CI + H 3 C-Si- HC|» -Si-CH2-Si- (31)
Secondly - at higher temperatures (T > 380 °C) the polysilane-to-

poly(carbosilane) transformation becomes the dominating process (Eq. 28).
From 180 to 350 °C building up of polycyclic polysilane network is typical.
Above 350 °C degradation of the polysilane skeleton due to an enhanced
cleavage of Si—Si bonds and generation of -Si-CH2-Si- sequences are
dominating. However, in contrast to the pyrolysis of methylpolysilanes, the
conversion of such chloromethylpolysilanes did not give any products
involving Si—H bonds. This fact should be related to condensation steps
consuming Si—H bonds (Scheme 22).
3.2
Transformation of the Amorphous Si-C Network into Poly-Crystalline SiC
3.2.1
Amorphous Covalent Ceramics
After a heat treatment of the precursor polymer up to about 800 °C a state has
been reached which is termed "amorphous covalent ceramics" or "ACC
phase" [133]. It is characterized by the following items [133-135]:
- The weight loss is more or less finished for the present, i.e., during a further
increase of the temperature only hydrogen - which is contributing only at a
negligible rate to the mass loss - (and in the special case of chlorine
containing polymers also traces of HCl) are detectable by mass spectroscopy
or gas chromatography up to the temperature range of about 1200 °C. Here,
reactions between free carbon and often unavoidable regions of silica start,
and once more weight losses are observable.
- Densities are - depending on the real composition and foregoing treatment
- in the range 2.2-2.4 g cm" 3 .
- Hints of crystallinity cannot be detected neither by XRD nor by TEM or
SAED (selected area electron diffraction).
- In the IR spectrum, the Si—CH2—Si wagging vibrations in the range of
1000-1100 cm" 1 which are characteristic for the existence of poly(carbos-
ilane) have disappeared completely.
- The 29Si MAS NMR signal shows only one single peak with a chemical shift
of about - 8 ppm, typical for a [SiC4] coordination. However, the linewidth
of this peak is large (about 35 ppm), i.e., the surrounding of the silicon
atoms is still irregular. Signals which could arise from residual Si-Si bonds
of the polymer are no longer observed.
- The 13C MAS NMR spectrum is characterized by a broad signal at about
18 ppm which is correlated with an irregular [CSi4] coordination. In the case
of systems with a C/Si ratio significantly larger than 1 [133, 134], a broad
peak around 135 ppm is also detected arising from C=C bonds.
- The spin densities measured by EPR are still relatively low (0.5 x 1019 cm"3
[135] - 1 x 1019 cm"3 [134]) compared to values obtained after a treatment
at higher temperatures (up to 1020 cm"3 at 1000 °C).
Beyond all doubt, the state of the "amorphous covalent ceramics" forms a
specific and, by knowing the preparation route, a more or less exactly
definable state which occurs as an intermediate during the transformation of
the silicon organic polymer into the final crystalline ceramics. This state
appears to be stable at room temperature for a long time. And, of course,
materials in this state are distinguished by specific properties different from
the state of crystalline ceramics.
However, the authors are reluctant to denote this state as ACC "phase".
The term "phase" implies - by its meaning in chemical thermodynamics - an
equilibrium state. It refers to a state which is unequivocally defined by the
fhermodynamic parameters like temperature and pressure present or, in
other words, a material constituting a real phase is characterized by
properties which are independent of the route leading to that phase. In the
case of the amorphous covalent ceramics, however, the properties will depend
strongly on the particular history of the material, on parameters like heating
rate, maximum temperature and length of the heat treatment, kind of
atmosphere during this treatment, etc. Therefore, it is wrong to speak about
the ACC state of SiC. In reality, a lot of different ACC states in the Si-C
system are possible, depending on the real formation route and characterized
by differences in properties like density, electrical conductivity, spin density,
etc.
Obviously, such a state may be termed a "frozen state", far from the
thermodynamic equilibrium. However, we cannot agree to a definition given
by Chawla [136]: "Noncrystalline or amorphous (...) ceramics are not really
solids but supercooled liquids.... Thus, noncrystalline ceramics such as glasses
may, in many respect, behave like solids but structurally they are liquids."
Without any doubt this statement is true of amorphous ceramics like the
usual silicate glasses which can be formed from melts by rapid cooling. In the
case of amorphous ceramics organized as covalent networks such a definition
does not correspond to the reality. This is especially obvious in the case of the
amorphous Si-C ceramics: not only does a silicon carbide melt from which a
supercooled state could arise not exist under normal pressure, but also the
chemical composition of the amorphous ceramics is far from that one of an (at
high pressure) imaginable SiC melt since the solid amorphous state is
stabilized by hydrogen (as well as by excess carbon, impurities etc.). The
relative stability of the solid amorphous Si-C network is due to the saturation
of the majority of "broken" (better: "not yet formed") Si—C bonds by H atoms
[134]. It is to be expected that under conditions where a fictitious melt could
exist such Si—H bonds are no longer stable. Thus it will be improbable to find
the most important structure units of the amorphous solid also in the
(unknown) structure of an (unknown) melt.
It is the authors' opinion that one has to concede that structures of

amorphous ceramics are also possible which are not related to structures of
corresponding supercooled liquids!
In this connection, however, it may be useful to discuss briefly the
correlations or similarities between the well-known structures of silicate
glasses (see e.g. [137]) derived by supercooling from a melt and structure
elements of an amorphous Si-C network.
In both cases the basic structure units are tetrahedra around silicon atoms,
the [SiO4] tetrahedron and the [SiC4] tetrahedron, respectively. By sharing
corners of the tetrahedra the corresponding networks are generated. In the
case of crystalline materials, the structure of pure silica requires that the
bridging oxygen atoms connect two tetrahedra [SiOSi] in every case, whereas
the silicon carbide structure needs the connection of four tetrahedra by one
bridging C atom also forming tetrahedra around the C atoms [CSi4].
In order to prepare silicate glasses it is useful to introduce additionally
so-called network modifiers MOX (x < 2, typical M: alkaline or alkaline earth
metals) beside the network former SiO2 into the system. From this a changed
structure results characterized by the occurrence of broken Si—O bonds. That
means beside bridging oxygen atoms non-bridging oxygen atoms also occur,
and the modifier cations are coordinated by such non-bridging oxygen atoms
preferentially. In other words, the formation of an amorphous glass structure
is kinetically favored with respect to possible crystalline structures by the
existence of modifier cations as well as non-bridging oxygen atoms, because
they complicate and impede the formation of a long range order of the silica
tetrahedra, even in the cases of stoichiometric silicate systems. In particular,
the regularity of such local oxygen polyhedra around the cations in the glass
structure may play an important role for the crystallization tendency of the
glass [138].
In the case of amorphous Si-C networks first of all the covalently bonded
hydrogen atoms take over the - structure dominating - role of the modifier
cations discussed in ionically bonded silicates. Hydrogen atoms occurring in
the Si-C system modify the network since they occupy one or more corners of
the tetrahedra around Si and C atoms, respectively. They act as non-bridging
atoms like the non-bridging oxygen atoms in silicate glasses. (Of course,
beside hydrogen other low-valent atoms like oxygen, nitrogen or boron may
act as network modifiers in this system. Their influence as well as the effect of
excess carbon will be discussed in the next section. A non-bridging
tetrahedron corner may also be formed by a dangling bond instead of a
hydrogen atom.)
However, there are important differences between the oxide and carbide
networks of silicon.
In order to crystallize silicate glasses it is not necessary to remove the
modifier cations since they can interact with more than one oxygen
tetrahedron due to their ionic bonding character. Thus, these cations act as
connecting links between the non-bonding oxygen atoms. First of all one deals
with a kinetic problem, and often a previous phase separation process is
needed to get crystallizable components.
We will see in the next section that phase separation processes, although
occurring in Si-C systems, are not sufficient to initiate crystallization processes
in the carbide system, but that we also have to remove the hydrogen atoms as
"network modifiers". Due to its monovalently covalent bonding character the
hydrogen atoms cannot act as connecting links to neighboring tetrahedra.
Thus only the removal of the hydrogen atoms can lead to crystallizable
structure units. However, that means simultaneously that the crystallization
process in such polymer derived ACC systems is always combined with
changes in the chemical composition. Therefore, there is no kind of
reversibility between amorphous and crystalline states - in contrast to most
silicate glasses.
Hitherto we have discussed the formation of amorphous covalent ceramics
only on the basis of polymer derived materials. In Sects. 2.2 and 4.2.2.2, thin
amorphous, hydrogen stabilized SiC layers (a-SiC:H) are also considered
which are formed, first of all, by gas phase processes (CVD, PVD). They
represent another type of amorphous covalent ceramics. And though it is not
expected that properties of such layers agree completely with those of the
polymer derived ACC, the basic ideas of stability and transformability of the
ACC state discussed above should be transferable to this type of amorphous
covalent ceramics, too.
3.2.2
Nucleation and Crystal Growth
The mechanisms of the transformation from polymer derived amorphous

networks of different Si-C-(H-O-N) systems to crystalline SiC ceramics as well
as approaches to control these mechanisms have already been investigated for
more than one decade. (Of course, also more complex Si-C systems containing
further heteroatoms like B, Ti, Zr, and Al are interesting today due to technical
importance of the resulting ceramics, but we want to focus our discussion on
systems resulting from simple SiC precursor systems like, first of all,
polysilanes and polycarbosilanes, without excluding polysilazanes and poly-
siloxanes.) Beside papers already quoted in the last section [133,134], there are
a lot of publications of the same and other groups engaged in this field, e.g.,
[139]. Also some results from the authors' groups concerning the crystalli-
zation of polymer derived SiC have been published [140-146]. However, one of
the most important papers concerning this issue is a paper summarizing a
series of investigations of Monthioux and Delverdier [ 147].
In the majority of polymer systems designed for an SiC synthesis, the atom
ratio C/Si is significantly >1. Therefore, in the crystallization model described
by Monthioux and Delverdier [147] the excess carbon is considered to play a
very important role. By means of HRTEM it was observed that more or less in
all the systems investigated the formation of the so-called "Basic Structure
Units" (BSU) of the free carbon - i.e. the formation of small stacks of only a
few polyaromatic layers - seems to be the first step for the following nucleation
mechanism of SiC. Such BSUs (graphenes) have lateral extensions in the order
of 1 nm, are laterally saturated by hydrogen atoms and piled up in a
turbostratic form. During a further heat treatment by cleaving hydrogen the

BSUs start to be linked to one another and - depending on the amount of free
carbon - to form disturbed graphene layers or, finally, even a network of free
carbon. Simultaneously with this edge-to-edge association of the BSUs, the
first SiC nuclei are observable. From this, Monthioux et al. conclude that a
kind of heterogeneous nucleation takes place. Due to similarities in the
structures (or, at least, in the symmetry) between graphene and /? SiC (111) the
BSUs are considered to act as a template for the formation of SiC nuclei. In
other words, the first main events in the crystallization process are happening
in the sequence: nucleation of BSUs —> edge-to-edge association of the BSUs
by forming distorted graphene layers —> nucleation of SiC crystals.
It is not surprising that the starting temperature of that process chain also
depends on the type and amount of further heteroatoms present in the
amorphous network, since the activation energies of the nucleation as well as
of the diffusion will rise in this case. Monthioux et al. summarized the results
of their observation in a simple scheme by including also those temperature
ranges where degradation processes take place in the case of the more complex
systems Si-C-N, Si-C-0 and Si-C-N-O. In the case of Si-C systems results
Table 1. Survey of the temperature dependence of the main physico-chemical events on the
thermal evolution of some polymer derived ceramic materials (after [147])
chemical heat-treatment temperature in 'C

system
C/Si 900 950 100(1 1100 1200 1250 1350 1400 1450 1600
Si-C 1.1 Cr
limn dill
1.4 (BSU) d/Cr of the crystallized
cei •a mic
2.1 BSU (d/Cr) d/Cr
Si-C-N (BSU) d/Cr

d o n ain
of (h< : amorp Mil". BSU d/Cr degradation I

Si-C-O
1 domain 1
Cera mic
Sl-C-N-O BSU [cr|
Cr SiC crystal nucleation
d/Cr edge-to-edge association of BSUs into distorted graphene stacks
BSU BSU nucleation
|H thermochemical degradation of the heteroatom- containing intergranular phase
concerning C/Si ratios of 1.4 and 2.1 were listed in the scheme which is given
in Table 1.
The first row in this table (C/Si =1.1) was added by the present authors and
resulted from [144-146]. By investigations which will be described in detail in
the following section, we have been able to show that in amorphous systems
derived from chlorine containing polysilanes (see Sect. 2.4.2) - where a nearly
stoichiometric C/Si ratio becomes possible - obviously a SiC nucleation is
occurring already in the temperature region of only 900 °C without a
preceding formation of BSUs!
Which phenomena give hints for another nucleation mechanism besides the
one described by Monthioux et al.?
First of all, the paper of Monthioux itself it should be mentioned. He
considers the dependence of the temperature of the first BSU occurrence on
the carbon content of the system and finds an inverse correlation between
these two parameters: the lower the carbon content is, the higher a
temperature of BSU formation is observed. This can be understood by taking
into consideration that the driving force for the BSU formation will increase
with increasing carbon content as well as that the diffusion paths during that
formation will decrease. Therefore, already Monthioux seems not to be
absolutely sure if in the case of the C/Si ratio of 1.4 a foregoing BSU formation
really occurs and gives only a cautious reference to a possible BSU formation
by setting it in parentheses (Table 1). In another paper of this group [148] it is
explicitly reported that after a treatment at only 900 °C the formation of BSUs
is not yet observable.
By using data given by Monthioux a rough estimation of the temperature
for a possible BSU formation in case of a C/Si ratio of about 1.1 leads to a
temperature of about 1050 °C. That is more or less compatible with the
following own results [146]:
- Investigations of the atmosphere during pyrolysis and crystallization
processes via gas chromatography show that hydrogen - cleaved also by
the edge-to-edge association of the BSUs - is found up to temperatures
exceeding 1000 °C.
29
- Si CP MAS NMR signals which need the presence of hydrogen atoms
disappear at T > 1100 °C.
- The content of free carbon detectable by hot gas extraction (LECO RC-412)
begins to increase only above T = 1100 °C.
- EPR investigations reveal a characteristic change in the kind of the
dominant signal. In the temperature range 1050-1100 °C a g-value of 2.003
is observed which is characteristic for radicals of carbon (or also -
g = 2.0035 - for radicals of silicon atoms bonded to carbon [149]). At
T > 1050 °C this signal has disappeared and a signal with a g-value of 2.005
becomes dominant corresponding to Si radicals. It means that, obviously,
C—C bonds were formed predominantly by combining C radicals (or that
even Si—C bonds were cleaved to form C—C bonds.)
On the other hand, there is some experimental evidence for a temperature
of maximum nucleation of silicon carbide at about 900 °C in our ACC system
with a C/Si ratio of 1.1 [144-146] (which would also be in agreement with
former 29Si MAS NMR studies of Soraru et al. on pyrolyzed polycarbosilanes
[150] revealing that the linewidth characterizing the disorder in the silicon
environment is increasing only up to 840 °C, whereas a treatment at higher
temperatures leads to a more regular silicon coordination as a first step to the
SiC nucleation.)
To investigate the nucleation behavior of our system we applied methods
and theories usually used for analyzing the crystallization processes of glasses
[151-153].
By starting from an ACC system characterized by a standardized pre-
treatment up to 700 °C (heating at 2 °C min" 1 to 700 °C and holding for 1 h
under an argon atmosphere) a series of samples was firstly pyrolyzed for 8 h at
different temperatures in the range of supposed nucleation (800-1100 °C) and
finally all the samples were held for 20 min at the same temperature
T = 1500 °C in order to let the nuclei grow which had been formed before at
the lower temperatures. These samples were analyzed by XRD.
Crystallite sizes dXRD determined by measuring the line broadening in XRD
reveal a distinct dependence on the nucleation temperature. We find a
significant minimum (dXRD, 900 = 8 nm) for samples nucleated at 900 °C,
whereas these mean crystallite sizes increase at lower as well as at higher
nucleation temperatures (dXRD> 8Oo = 16 nm, dXRD; 1100 = 12 nm). It means
that at T = 900 °C the most nuclei were formed, with the consequence that
under the final uniform crystallization regime at 1500 °C the many nuclei
formed at 900 °C were able to grow only to a lower degree until the crystallites
touched one another. The crystallizable material has been consumed, and the
crystal growth is stopped earlier, than in the other cases where a less number
of nuclei formed at lower or higher temperatures.
SAXS investigations (which were performed by A. Hoell and R. Kranold at
the University of Rostock, Germany) lead to similar conclusions. By that
method sizes of heterogeneities ("particle sizes") as well as their relative
frequency per volume ("relative particle density") and the relative volume of
all particles ("volume fraction") can be determined. An advantage of this
method is the feasibility to detect particles with dimensions already in the
range of the nuclei themselves, i.e., after the nucleation no crystal growth at
higher temperatures is necessary for detection as is needed in the case of XRD.
Therefore, samples used for these investigations were held only at the
nucleation temperature (900, 1000, 1100 °C), but with additional variations of
the nucleation time (1, 2, 4, and 8 h).
- By increasing nucleation time one can observe that the particle sizes dSAXS
remain nearly constant (dSAXS; 900 =1.2 nm) for the system nucleated at
900 °C, whereas at 1000 °C a slight (dSAXS> 1000=1-5 nm -» 2.2 nm) and at
1100 °C a significant (dSAxs, iio<>2.1 -> 4.3 nm) growth of these sizes is
observable; i.e., obviously, the temperature of only 900 °C is not sufficient to
activate the growth of particles, but only to form nuclei.
- The relative particle density is significantly the highest one for T = 900 °C
and increases with nucleation time (especially by going from 1 to 2 h) only
up to 1000 °C. At 1100 °C this relative number of particles per volume

remains constant and has a value which is (at 8 h) about 15 times smaller
than the corresponding value for T = 900 °C and about 5 times smaller than
that for T = 1000 °C; i.e. obviously, the conditions for particle nucleation
are at an optimum at T = 900 °C.
- The volume fraction, however, most significantly increases with time for
samples treated at 1100 °C, i.e. although in this case the number of particles
remains low, since the temperature region of optimum nucleation has been
passed too rapidly, at t = 8 h the volume fraction of formed particles is the
highest one due to the particle growth in the amorphous matrix becoming
important first in this temperature region.
These interpretations are supported by DTA measurements already
reported in detail [144, 145]. By using the model proposed by Marotta et al.
[151] and further developed by Ray et al. [152, 153] the exothermic peak of
crystallization, i.e. of the optimum crystal growth in the amorphous matrix,
has been used to investigate the kinetic of crystallization. For our amorphous
SiC system this peak is observable in the range 1050-1130 °C, and its exact
position T p as well as its height (5Tp depends on parameters like the
temperature of nucleation treatment and the heating rate during the DTA
experiment.
By considering the peak height <5Tp as a function of the nucleation
temperature a distinct maximum can be observed at T = 900 °C; i.e. at this
temperature the most nuclei had been formed which can grow and give a
contribution to the energy from which the exothermic crystallization peak
during the DTA experiment arises!
Further information on the kinetics of the crystallization process can be
obtained by this method by variation of the heating rate during the DTA
experiment. By increasing the heating rate Q (5 K min" 1 up to 40 K min" 1 ) a
systematic shift of T p to higher temperatures is detectable.
Applying the Kissinger equation [154]
ln(Q/Tp) = -(nEA!Cr)/(RTp) + lnc (32)
with EA;Cr as the activation energy for the growth of crystallites in the
amorphous matrix, n as a parameter correlated with the Avrami exponent
(here n was determined to be equal 1), R as gas constant and c as a value
independent on Tp, one can use that observable shift of T p = f(Q) to estimate
the activation energy.
Samples thermally treated at the optimal nucleation temperature of 900 °C
give an activation energy in the range of optimal crystal growth (about
1100 °C) EA,Cr = 576 ± 50 KJ mol" 1 . If the same experiment is performed
without nucleation pre-treatment at 900 °C the activation energy (in the same
temperature range of about 1100 CC) is found to be EA- = 667 ± 20 KJ mol" 1 .
Our interpretation of this difference is the following one.
Both processes, nucleation and crystal growth, are processes requiring an
activation. However, the activation energies for the two processes are different.
In the first case we can be sure that the growth of nuclei formed earlier will be
the dominant mechanism responsible for the measured activation energy, i.e.,
it is more or less the real activation energy for the growth of crystallites from
an amorphous matrix (but also influenced by further nucleation processes
which still take place at T > 900 °C!).
In the second case our result is equally influenced by both processes,
nucleation and crystal growth. Of course, it does not make sense to try to
attribute the difference between these two values to the real activation energy
for the nucleation process alone because both processes are "smeared" and
overlapping in the temperature range. However, obviously the activation
energy for nucleation is significant smaller than the one for crystal growth.
This is in agreement with former results of Soraru et al. [133] who determined
an activation energy for the densification of SiC ACC in the temperature range
900-1000 °C - i.e., in a range where the nucleation forms the dominating
process, but also already overlaps with growth processes! He found
343 KJ mol" 1 .
An activation energy in the range 550-600 KJ mol" 1 reflects the mecha-
nisms responsible for the crystallization. Obviously, at the interface between
the growing nucleus and the amorphous matrix, reconstructions are necessary,
which require, on average, the cleavage of more than one covalent bond (e.g.:
Si-H: 323 KJ mol" 1 , C-H: 416 KJ mol" 1 , Si-C: 306 KJ moP 1 [155]). This means
that the cleavage of chemical bonds connected with the formation of radicals is
the process dominating the crystallization. That is absolutely in agreement
with our EPR investigations mentioned above, where the maximum of the
radical concentration was detected after a treatment at exactly the same
temperature (1050 °C) where the exothermic DTA peaks begin to form. Taking
into consideration that the EPR signal from samples treated at 1050 and
1100 °C are dominantly caused by carbon radicals, it does not seem to be
misleading to interpret the formation of these radicals as the step dominating
the growth of SiC crystals from the ACC.
A similar value of the activation energy for the crystallization of SiC like the
second one mentioned above was found by Yoshi [156] investigating the
crystallization behavior of amorphous Si!_xCx films prepared by RF sputtering.
He gives a value of 685 KJ mol" 1 . In that connection, it may also be interesting
to mention that the activation energy for self diffusion processes of Si and C in
SiC was found to be still higher: 8 eV = 772 KJ mol" 1 [157].
The DTA measurements reveal a further - somewhat smeared - exothermic
peak in the temperature range from 1390 to 1510 °C. Obviously, this peak
arises from processes of re-crystallization. By XRD a drastic increase of the
crystallite sizes is observable in the same temperature region. Thus, when the
ACC is consumed by the crystallization process, a new mechanism starts to be
responsible for the development of the microstructure of the final SiC
ceramics. A further increase of the crystallite size becomes possible only by re-
crystallization, i.e. Ostwald ripening. The activation energy for such processes
determined by the Kissinger equation in the way mentioned above was found
to be significantly lower: EA; recr = 381 ± 20 KJ mol" 1 .
All the results reported hitherto refer to our ACC system with a C/Si ratio of
about 1.1 which leads to a nearly stoichiometric SiC ceramic. Of course, in spite
of a careful handling of this - up to about 1000 °C also chlorine containing -

precursor system, a contamination with traces of oxygen were not avoidable in
every case due to states of very high reactivity during the whole transformation
process. However, in every case we analyzed the oxygen content of our
samples in various states of the transformation process, and samples with a
significant oxygen content were rejected since it is known, especially from the
thorough paper by Monthioux and Delverdier [147] discussed above, that
every kind of heteroatom will shift the temperatures for nucleation and
crystallization to higher temperatures.
Therefore, we will also have a quick look at the crystallization behavior of
ACCs formed by SiC precursors containing heteroatoms like polycarbosila-
zanes or polycarbosiloxanes which are also listed in Table 1. In general, we
can see that the nucleation steps for SiC take place at temperatures where
carbon BSUs have already formed. Therefore, the crystallization mechanisms
using such BSUs as templates for SiC nuclei proposed by Monthioux et al.
seem to be plausible and will not be discussed further. However, in the
authors' opinion, the theoretical approach of Monthioux et al. to understand
the crystallization behavior by generally correlating the formation of BSUs
and the nucleation of SiC implies a principal error since the driving force for
the formation of BSUs will really decrease with decreasing C/Si ratio, whereas
the tendency for the formation of SiC nuclei will be the highest if the C/Si ratio
approaches unity!
The model of Monthioux et al. describes the crystallization behavior of SiC
also for more complex precursor systems Si-C-N, Si-C-O, and Si-C-N-O. In the
case of nitrogen-containing systems it has been observed [158] that, beside the
"bulk" mechanism based on BSUs discussed above another mechanism, a
"surface" mechanism may also occur. Especially on surfaces of fibers derived
from polycarbosilazanes, needle-like silicon nitride crystals are formed under
a nitrogen atmosphere via gas phase processes at T > 1600 °C.
Similar phenomena were thoroughly investigated by Kleebe et al. [159-161].
Starting from a polyvinylsilazane precursor they found that, obviously, gas
phase processes begin to play a role for the crystallization of the ACC on inner
surfaces already at T > 1250 °C. A very impressive proof for these phenomena
gives Fig. 2. It shows a SEM micrograph of an inner crackwall of SiCN
annealed at 1540 °C in N2. The inner surface is covered by a lawn of small SiC
crystallites with idiomorphic a-Si3N4 single crystals grown on top.
It is worth to mention that under the same conditions (1540 °C, N2) leading
to surface crystallization via gas phase processes, Kleebe et al. were able to
show by TEM and XRD that, within an amorphous SiCN matrix, globular
inclusions also exist where crystallization has taken place and where
crystallites of graphite, silicon carbide, as well as silicon nitride were found
side by side. However, the thermodynamic stability of this system vanishes at
higher temperatures [162]. At T = 1600 CC silicon nitride has disappeared
and, e.g., the 29Si NMR spectrum reveals only [SiC4] (and no longer [SiN4]
which had been detectable at T < 1500 °C.)
All the papers cited hitherto demonstrated that the crystallization tendency
is shifted to higher temperatures if heteroatoms like oxygen or nitrogen are
Fig. 2. Inner surface of SiCN ceramic with large Si3N4 single crystals and small SiC crystallites
grown by gas phase reactions under annealing in N2 at 1540 °C. (Courtesy of H.-J. Kleebe,
University of Bayreuth, Germany)
participating in the formation of the Si-C amorphous network. Of course, there

are also additives which may attack this network at temperatures before the
crystallization of SiC has started and which may form crystalline reaction
products at relatively low temperatures. For example, in our group the
pyrolysis of polysilane under addition of tungsten [163] and titanium [164] has
been investigated. In these cases the crystallization process starts with the
formation of silicides or carbides of the added metals. By XRD at T = 700 °C
crystalline phases (TiSi, Ti5Si3) were already detectable in the case of Ti
addition, whereas the addition of W led to the formation of crystalline W2C at
1000 °C.
It is to be expected that such phases may also control the later
crystallization of SiC by heterogeneous nucleation. Such effects have been
described in [150, 165, 166] concerning the crystallization of Si-C-(O)
precursors containing metals like Ti, Zr, Al, i.e., precursor systems from
which different types of Tyranno fibers are formed [167]. In the case of the
Al-containing system it was observed that - under the given unusual
conditions - SiC polytype 2H is stabilized by A12OC which is isostructural to
SiC • 2H.
A final comment on the crystallization of SiC from polymer derived ACCs is
that, more or less, in all papers regarding this problem it is reported that
silicon carbide is formed as jS-SiC, i.e., the polytype 3 C. Our group followed
the polytype development of SiC formed by pyrolysis of polysilane in
dependence on the pyrolysis temperature [168] and found that, indeed, up to
1800 °C P-SiC dominates the XRD, Raman, as well as 29Si NMR results, but
that especially in the case of pyrolysis temperatures in the range 1600-1700 °C
non-negligible parts of polytypes of a-SiC also occur which have disappeared
again at T = 1800 °C.
4
Alternative Routes to SiC
4.1
Routes Using Phytogenic Material
4.1.1
Routes Starting from Sugar
The classical route to SiC, the so-called Acheson process, has recently been
described thoroughly by Schwetz [169]. It is well known that high-purity
quartz sand and petroleum coke are used to get silicon carbide corresponding
to the summarizing equation for this strongly endothermic carbothermal
reduction process (ARH298 = 618.5 KJ mol" 1 ):
SiO2 + 3 C SiC + 2 CO (33)
and that high temperatures (typically exceeding >2000 °C) are needed.
If we use the term "alternative routes to SiC" here, we want to exclude this
Acheson process as well as the main topic of this paper, the polymer route!
By considering the thermodynamical data of this carbothermal reduction it
becomes obvious that the temperatures used in industrial scale applications
are significantly higher than the one needed to get a negative Gibbs energy for
the reaction. (ARG becomes zero at about 1800 K.) This is not to mention that
in the case of such an equilibrium where one of the products may escape in
open systems, the reaction would take place also already at lower tempera-
tures. Thus, the need for such high temperatures is predominantly due to
kinetic reasons.
The mechanism of this carbothermal reduction has been intensively
investigated [170, 171] and a scheme of consecutive partial reactions has
been proposed revealing that the reaction rate is controlled by diffusion and
gas phase processes. Therefore, it can be concluded that the reaction rate may
become the higher the finer the distribution of the solid educts is. That is, if we
are able to start from a colloidal dispersed mixture of silica and carbon, it is to
be expected that the reaction temperature will drop as well as that finer SiC
particles can be obtained.
Such considerations have led to a process forming finely dispersed and pure
SiC powders by starting from a solution of sugar (saccharose) in silica sol
[172]. After drying the solution and pyrolyzing the sugar, an intimate nano-
scaled mixture of silica and carbon results which may be transformed into SiC
by pyrolysis under an inert atmosphere at T < 1800 °C.
In the last decade there have been some efforts to optimize this process in
order to drop the reaction temperature and to get nano-scaled SiC powders
without agglomeration and contamination [173-175]. Our group was also
engaged in this field [176, 177]. We were able to show that it is feasible to
control the particle sizes by controlling the gas pressure during the reaction as
a function of the reaction temperature. In order to minimize the particle
growth, the carbothermic reduction is suppressed during the heating phase by
0 150 nm
Fig. 3. TEM micrograph of a SiC powder prepared from silica sol and sugar at T = 1800 °C.
(Courtesy of Ch. Oestreich, TU Freiberg, Germany)
gas pressure; only when the intended reaction temperature has been reached is
the pressure lowered down to a vacuum of about 20 Pa with the consequence
that the reaction takes place very abruptly. Thus, nano-scaled SiC powders
characterized by specific surface areas of 25-30 m2 g~x were feasible at
reaction temperatures of 1500-1800 °C. Figure 3 shows a TEM micrograph of
such a powder synthesized at 1800 °C. Stacking faults inside the monocrys-
talline particles of /?-SiC are distinctly observable.
4.1.2
Routes Starting from Silica-Containing Plants
It has already been well known for more than two decades that it is feasible to
produce SiC powders and/or whiskers by pyrolyzing rice husk [178] under an
inert atmosphere, since a considerable content of colloidal amorphous silica is
incorporated in husks and straws of this vegetable. It means that nature by
itself is supplying a similar intimate mixture of carbon-containing molecules
(cellulose) and colloidal silica like those artificially made mixtures we
discussed in the last section. Thus, also in this case, silicon carbide may be
formed by a carbothermal reduction process. This process starts in the
temperature range 1100-1400 °C!
Although this process is already used industrially for producing SiC
whiskers [179], recently investigations have been performed with the aim to
control the shape of the produced particles - globules or whiskers - by process
parameters [180, 181].
In the authors' group similar investigations have been performed using
silica-containing plants available in Germany. We chose the so-called "horse-
tail" (Equisetum arvense), a weed-like vegetable, which traditionally had been
Fig. 4. SEM micrograph of a leaf of horse-tail (Equisetum arvense). (Courtesy of M. Ade, TU

Freiberg, Germany)
used for cleaning the surfaces of pewter (tin vessels, table-ware, etc.). That
cleaning effect is due to the abrasive influence of small silica-like inclusions,
preferentially inside the tops of the needle-like "leaves" arranged like twirling-
sticks. Figure 4 shows a SEM micrograph of a top of such horse-tail leaves. By
EDX the white points reveal a significantly higher amount of silicon than the
residual parts of the plant.
We found a silicon content of about 5-6 wt% in dried material from the
whole plant. That content increases up to 8% by analyzing only the leaf tops.
Of course, there are also various mineral impurities (mainly Ca, K, S, Cl) in the
dried vegetable material (up to about 10 wt%), whose type and amount may
depend significantly on the place from where the material arises. However, the
majority of such impurities can be washed out before pyrolysis.
After pyrolyzing under Ar at 1400 °C a nano-scaled SiC powder is obtained,
but the XRD pattern still reveals a significant background arising from residual
amorphous silica. A pyrolysis at 1550 °C, however, yields a powder charac-
terized by the distinct XRD pattern of /3-SiC and residual free carbon. The
mean particle size derived from the line width of the SiC diffraction peaks is
about 30 nm.
4.1.3
Routes Starting from Wood
In the last decade there has been much activity concerning the use of biological
materials while maintaining their structures for producing advanced ceramic
pyrolysis carhruni/ed
Ih 70 C ~~^ 18(WC41i N , — preform —
Sl-eas nrui ration

-*- SiC- ceramics
lMO'C •Jli vacuum
Tflildl nfiltration
-» SiSlC-ceramics —p- etchinn - Sit" ceramics
*" L600'( 4h argon ~~
Fig. 5. Processing scheme of manufacturing different types of cellular SiC ceramics from
wood (after [191])
as well as composite materials with a biomorphic microstructure. Biological

structures are characterized by complex shapes of a hierarchically built
anatomy leading typically to an anisotropic behavior of the organic template
as well as of the resulting ceramic material. Beside intensive investigations to
produce ceramic fibers (SiC, B4C, A12O3, TiO2, ZrO2) by using biological fiber
templates [182-186] there are some groups engaged in the transformation of
wood into biomorphic SiC ceramics [187-189]. The current importance of this
problem is emphasized by the fact that, e.g., at the last German meeting
"Composites and Compound Materials" (October 1999 in Hamburg) alone
four papers were presented concerning the formation of silicon carbide
materials from natural wood or wood products [190].
The scheme of the typical processing for the transformation of wood into
SiC ceramics is given in Fig. 5 [191].
From that scheme it becomes evident that - in contrast to the processes
discussed in Sects. 4.1.1 and 4.1.2 - a carbothermal reduction process no
longer takes place to form SiC. (Of course, such processes are also feasible. Ota
et al. [187] used TEOS for infiltration after carbonization, but the process
described here is more effective.) After the carbonization of wood a direct
reaction between carbon and silicon is used at which silicon is infiltrated into
the porous carbonized wood system either by a silicon melt through capillary
forces or by gas phase processes. Microstructures of carbonized as well as of
Si-infiltrated wood are given in Fig. 6a, b. Figure 6a shows the structure of
beech after a pyrolysis at 1800 °C (4 h) under a nitrogen atmosphere; Fig. 6b
gives the same structure after the subsequent infiltration of silicon melt at
1600 °C under vacuum. The properties of the resulting anisotropic SiC
materials are described thoroughly in [192].
4.2
Silicon Carbide via Gas Phase Processes
4.2.1
Single Crystals
Due to its wide band gap, silicon carbide is also an interesting semiconductor
for high-temperature applications [193]. In order to get monocrystalline
£. -A •
Fig. 6a, b. The microstructure of beech: a after carbonization; b after a subsequent silicon
melt infiltration. (Courtesy of H. Sieber, University of Erlangen, Germany)
wafers well-known technologies usually applied for the wafer preparation of

silicon or various AIII-BV semiconductors may not be adopted since the
seeded growth of such materials is starting from melts. In case of silicon
carbide, however, a congruent melt does not exist under technologically
relevant pressures. Therefore, the growth of monocrystalline SiC boules is
performed by physical vapor transport via seeded sublimation originally
reported by Tairov and Tsvetkov [194]. At temperatures of 2200-2500 °C
gaseous species like Si2C, SiC2, and Si exist. They are responsible for the mass
3C-SiC 2H-SJC
•W A
B
4-H-SiC 6H-SIC
Fig. 7. Scheme of the structures of four SiC polytypes by comparing the views on (110) (3C)
and (1120) (2H,4H,6H), respectively. For emphasizing the differences in stacking sequence
in the direction [111] and [0001], respectively, a sequence of bondings has been marked
transport across an inert atmosphere (Ar, He, or N2) from the SiC source at
higher temperature to the seed at lower temperature in the course of which
reactions of the gaseous species with the crucible wall of carbon may take
place, too.
Recent reviews on the growth of high-purity SiC single crystals and their
properties have been given by Miiller et al. [195] and Augustine et al. [196].
Today, wafer diameters up to 50 mm are feasible on an industrial scale for the
polytypes 4H and 6H; the density of micropipe defects can be reduced to less
than 100 cm" 2 (down to about 1 cm" 2 ).
To illustrate the meaning of the polytype symbols used here, Fig. 7 shows
the structures of the simplest four SiC polytypes 3C (/?-SiC) as well as 2H, 4H
and 6H (polytypes of a-SiC). They are distinguished by the stacking sequence
of the layers formed by [SiC4] tetrahedra.
4.2.2
SiC layers
4.2.2.1
Monocrystalline Layers and Surfaces
In the last section the growth of single crystals has been described. In this case
the aim is to obtain crystals of uniform polytypes. On the other hand,
remarkable differences in physical properties of the different polytypes may
occur (see also Table 3): e.g. the cubic polytype 3C has a band gap of about
2.3 eV, in the case of the polytype 4H; however, this band gap is about 3.2 eV.
It has already been mentioned that this band gap predestines SiC for electronic
devices at high temperatures, but also for high power and high frequency
operations and devices operating in harsh environments [197]. Therefore, it
may be of interest to "tailor" this band gap for special applications. That is
feasible by the formation of so-called superlattices, i.e., by forming exactly
defined sequences of monolayers of different polytypes.
Therefore, growth conditions are needed other than the ones described
above to make possible controlled changes in the nucleation conditions in the
gas phase process. Such changes become feasible by molecular beam epitaxy
(MBE) [198]. Here the concentrations of gaseous silicon and carbon,
respectively, may be controlled independently by two separate electron beam
sources in an UHV system, and are simultaneously analyzed by mass
spectroscopy. A scheme of such equipment is given in Fig. 8. The feasibility for
reflection high-energy electron diffraction (RHEED) allows us to observe when
a monolayer of a defined polytype has completely formed. By changing the
partial pressures of silicon and carbon, respectively, as well as the temperature
of the substrate (800-1000 °C), it becomes feasible to control the nucleation of
a definite polytype independently of the polytype of the substrate [199].
In this way, superlattices composed of definite monolayer sequences of
different polytypes may be formed by MBE [200, 201]. Figure 9 shows a cross
sectional TEM micrograph of such a structure consisting of alternating
sequences of 6H and 3C layers.
heater (I650K)
RHEED screen
mass-s peel rom CU

pressure p< 10 Pa
UHV-putnping
Fig. 8. Principle of the MBE equipment for SiC. (Courtesy of A. Fissel, University of lena,
Germany)
no G. Roewer • U. Herzog • K. Trommer • E. Miiller • S. Friihauf
6H-S1C substrate
1.3C-SiC layer
1. SH-SiC layer
2. 3C-SiC layer
2. 6H-SiC layer
Glue
Fig. 9. Cross sectional transmission electron micrograph of a 3C/6H/3C/6H-SiC(0001)

heterostructure grown at 1500 K. (Courtesy of A. Fissel, University of Jena, Germany)
The problem is complicated by the circumstance that depending on

temperature and pre-treatment of the substrate, the surface may be recon-
structed in different ways, i.e., different superstructures are observable by
RHEED [e.g. on SiC(OOOl): (3 x 3), (2 x 2), and (V3 x V3)R30°], which may
influence the nucleation of the epitaxial layer. Recently, such superstructures
on SiC surfaces have been analyzed by LEED and modeled [202, 203].
4.2.2.2
Amorphous (a-SiCH) and Polycrystalline Layers
In Sect. 3.2.1 the importance of hydrogen for a stabilization of the amorphous

state of SiC has been emphasized. Therefore, in order to prepare such layers
via gas phase reactions the use of hydrogen-containing precursor molecules is
advantageous. The chemistry of these processes has already been discussed in
Sect. 2.2 (SiC synthesis from small molecules). Equally the most important
CVD techniques for preparing such layers have already been mentioned.
The importance of these amorphous layers derives from their electronic
structure. There are no longer sharp bands characterized by a definite band
gap, but quasi-continuous changes in the density of states are observed leading
to differences between the "optical gap" and the "mobility gap". Thus,
interesting optoelectronic properties and applications are reported, e.g.,
photoconductivity and solar cells [204, 205], optical windows for solar cells
[206, 207], electroluminescence and light emitting diodes (LED) [207, 208], or
thermistors for IR sensors [209].
A control of the properties is feasible not only by the chemical way via
doping already mentioned in Sect. 2.2, but also by the degree of amorphicity
and crystallinity, respectively, varying from absolutely amorphous layers over
nanocrystals in an amorphous matrix to polycrystalline SiC layers [210-212],
or by changing the C/Si ratio varying from a-Si:H to a-C:H [213, 214].
By the way, although a dominant majority of papers concerning the

formation of amorphous SiC layers describes applications of CVD or PVD
techniques, there have also been some attempts to use the polymer route for
preparing SiC films. Starting from solutions of various polysilanes or
polycarbosilanes, frequently films are formed by spin-coating and pyrolyzed
under inert atmosphere [215-218]. Of course, such a procedure does not form
a part of this section "SiC layers via gas phase reactions". However, in this
connection it should be mentioned that polysilanes are also applied to form
films via evaporation, not only with the aim to build amorphous and/or
crystalline SiC films, but also to use special properties of the polysilane films
themselves, i.e. without a subsequent pyrolysis of these films. Such amorphous
films are characterized by non-linear optical effects [219, 220] and their
properties may be controlled by the uniformity of the orientation of polysilane
chains which is susceptible to epitaxial influences [221-223].
4.2.3
SiC Fibers and Whiskers
4.2.3.1
SiC Monofilaments
In the following section (Sect. 5) an overview will be given over "thin" SiC
fibers characterized by "small" diameters of one to few tens of a micrometer
and produced via the polymer route (in most cases). This kind of fiber is
applied as rovings consisting of some hundred up to more than a thousand
single filaments, so-called multifilament fibers, which may be woven and is
predominantly used for reinforcing brittle matrices (ceramics and glasses).
Beside this type of SiC fibers another type is commercially available (e.g. the
type SCS-6 of Textron), so-called monofilaments, characterized by "large"
diameters up to about 150 (im which are formed by CVD processes. For
producing such SiC monofilaments thin substrate fibers are needed. Thin
carbon or tungsten fibers are used for that purpose. In the simplest case, they
are coated by SiC using, for instance, the thermal decomposition of CH3SiCl3.
The fiber type SCS-6 consists of a more complex, graded structure charac-
terized by a carbon core, an inner zone of carbon-rich /?-SiC, an outer zone of
stoichiometric /?-SiC, and a surface coating of carbon. Structure and properties
are thoroughly described in the literature [136 (p 92), 224]. An overview over
some properties is given in Table 2. Predominantly these fibers are used for
reinforcing metal matrices.
4.2.3.2
SiC Whiskers
Whiskers are monocrystalline fibers of very small diameters (down to 0.1 um)
typically grown by gas phase processes. Due to their nearly ideal structure and
the absence of crystalline imperfections like grain boundaries or dislocations,
SiC whiskers may possess outstanding mechanical properties; for example,
Table 2. Properties of SiC fibers produced by chemical vapor deposition [224]
Manufacturer Trademark Composition Diameter Strength Strain to Young's

(um) (GPa) failure modulus
(%) (GPa)
Textron SCS-6 SiC on C 140 3.4-4.0 0.8-1 427

DERA Sigma SiC on W 100 3.4-4.1 0.8 400-410
tensile strengths up to 23.7 GPa [136 (p 104)] were measured, i.e., the order of
magnitude of the theoretical value (about 84 GPa) estimated from strengths of
chemical bonds is reached.
Recently Schwetz [169] gave an overview of the different mechanisms
forming SiC whiskers. One of the processes technically used for SiC whisker
production has been mentioned before in Sect. 4.1.2, the pyrolysis of rice
husks. This process counts among the type VS (Vapor-Solid mechanism) since
the dominating reaction responsible for the whisker growth is the second step
of the reaction sequence:
SiO, + C — SiO + CO
(34)
SiO + 2C — SiC + CO
A scheme of the process technology for making silicon carbide whiskers
from rice husks is given by Chawla [136 (p 103)].
A second mechanism, the Vapor-Liquid-Solid mechanism (VLS) has also
been investigated thoroughly and is used industrially today. A particular
aspect of this process is the role of the "liquid": small droplets of metallic
melts, e.g. of iron alloys which may exist as an impurity in the reactor or which
are added to the process, act "catalytically" by dissolving gaseous species of Si
and C in the first step. This solution becomes supersaturated with time and the
precipitation of SiC begins. By continual dissolution and precipitation the
whisker grows bearing the droplet like a "head" on its top.
In the authors' group the formation of SiC whiskers has been observed for
the pyrolysis of the chlorine-containing polysilanes discussed in Sect. 2.4.2.
The pyrolysis was performed under streaming argon at temperatures up to
1500 °C. At very low streaming rates (<2 1 min" 1 ) whiskers of /f-SiC were
formed at cooler surface regions within the furnace, obviously arising from
escaping oligomers of the polysilane. Figure 10 gives an impression of these
whiskers characterized by unusually high aspect ratios (up to 1500) and
diameters in the range 0.2-2 um. No indication of a VLS mechanism was
found. The growth rate was estimated to about 30 um min" 1 .
4.2.4
SiC Nanopowders
By following the decrease in the dimensionality of SiC products formed by

gas phase reactions, after 3-dimensional bulks, 2-dimensional layers and
Fig. TO. SEM micrograph of p-SiC whiskers grown during the pyrolysis of polysilanes at
low streaming rates of argon. The electron diffraction pattern inserted into the corner
reveals the monocrystalline structure and the growth direction [111]. (Courtesy of B. Ullrich,
TU Freiberg, Germany)
1-dimensional fibers, finally, we want to discuss shortly the formation of

(O-dimensional) nano-scale silicon carbide powders.
Also in this connection the use of gas phase processes is of high importance
since a diminution of coarse SiC powders by grinding is an extremely time and
energy consuming process due to the hardness of SiC. Therefore, a "bottom
up"-method is the best way to get SiC nanopowders. As a consequence,
however, in contrast to the problems discussed in Sects. 4.2.1-4.2.3, the
conditions of the SiC formation via gas phase processes must be changed in
such a way that no longer heterogeneous, but homogeneous nucleation
dominates the process.
A survey of current methods for producing such nanopowders has
recently been published by Klein [225]. The monograph focuses on
thermally activated Chemical Vapor Synthesis (CVS) by using the educts
tetramethylsilane, hexamethyldisilane, and hexamethyldisilazane. The
author discusses correlations between process parameters, on the one
hand, and particle sizes, structure, and purity of the powders formed, on
the other. The pyrolysis temperature was varied from 800 to 1300 °C. By
introducing a cooler (77 K) for shortening the growth times of the
particles and for separating the particles by thermophoresis, average
particle sizes down to about 6 nm were achieved corresponding to specific
surface areas up to about 350 m 2 g"1. Powder yields of about 30 g IT 1
were attainable.
There are various modifications of this technique, especially regarding the
kind of the activation of chemical gas phase reactions. The most important
ones are the activation by various types of plasma [226-228] as well as by
lasers [229-231]. In all these cases it is feasible to obtain nano-scale silicon
114 G. Roewer • U. Herzog • K. Trommer • E. MCiller • S. Fruhauf
carbide powders. However, often parts or, at least, traces of elemental carbon
or silicon are found, too.
One of the advantages of lasers for activating gas phase reactions is the
feasibility to control the dimension of the process zone by laser focus. In this
way very small regions may be achieved where nucleation and particle growth
happen. The same advantage is applicable to processes where the laser is used
for evaporating coarse ceramic material in order to get nano-scale ceramic
powders by re-condensation via homogeneous nucleation [232]. This
technique has also been used to prepare SiC nanopowder with yields up to
100 g h" 1 [233]. However, with this approach, traces of free silicon were also
detectable.
5
SiC for High Temperature Application
5.1
SiC Properties Related to Advanced Applications
In the above sections new routes to SiC have been thoroughly discussed in
terms of chemical synthesis of SiC precursors, transformation to amorphous
and crystalline SiC, as well as alternative routes to SiC. The intention of this
concluding section is to review the potential of SiC for high temperature
application resulting from new fabrication technologies, illustrated by a few
typical parts and components.
The main potential of application results from the excellent high temper-
ature properties of SiC. General properties of pure SiC and conventional
manufactured SiC materials are listed in Table 3 [234, 235]
High creep resistance coupled with low density as well as good oxidation
resistance of pure SiC and most SiC-based materials compared with highly
advanced metallic alloys predestine this ceramic for devices operating at high
temperatures, up to 1500 °C, in the fields of:
Table 3. Properties of pure SiC and conventional SiC-materials

SiC material •L decomp X c a Density ^ultim Young-
Wm"'K"' J kg-'K"1 K1 gem 3 GPa modulus
GPa
a-SiC 42.5 690 4.3-4.6 3.21 450-520

jS-SiC 2400 135 1205 4.5 3.21 550
Monolithic SiC ceramics
SSiC at 20 °C 110 4.9 3.12 410 410
at 1370 °C 50 410
HPSiC at 20 °C 130 4.9 3.21 650 450
SiSiC at 20 °C 150 4.9 3.04 360 360
SSiC - pressureless sintered SiC

HPSiC - hot pressed SiC
SiSiC - Si infiltrated SiC
- Aviation and aerospace

- Automotive construction
- Machinery construction
Despite the fact that the strength to density ratio of SiC is significantly
better than that of Ni-based superalloys or refractive metals, brittleness of
monolithic ceramics is a limiting factor for high temperature structural use of
SiC. That is why SiC is toughened by embedding reinforcing components like
fibers, whiskers, or platelets. Flaws do not lead to catastrophic failure as in
monolithic ceramics, and damage tolerance and reliability are improved, when
cracks are deflected and bifurcated at inner interfaces. The most effective
toughening mechanism based on interface debonding and fiber pull-out can be
achieved with long ceramic fibers. The fiber/matrix interface region plays the
key role for inelastic behavior of this kind of composites [236, 237]. Preferred
reinforcing components for SiC matrices are carbon fibers as well as SiC-fibers,
-particles, -whiskers, and -platelets. That means SiC is of course not only an
important matrix material; it is also applied as reinforcement material in other
brittle matrices like alumina, zirconia, and silicon nitride [238-243] as well as
in glasses and glass-ceramics [244-246]. The reinforcement of metals with long
or short SiC-fibers is also common, yields stiffening - which is necessary for
light metals at higher temperatures - and is especially used for alloys based on
titanium or aluminum [247, 248].
High hardness and chemical resistance of SiC also at raised temperatures
yield good wear characteristics and lead to application of SiC in brake systems
as will be described in more detail below.
5.2
Fabrication of SiC-Based Fibers and Structures
Starting from SiC powder made, for instance, by the well known Acheson
process, the conventional processes for SiC components are sintering and hot-
pressing (for more details see [235]). Since the lack of toughness of monolithic
SiC is going to be compensated by the fabrication of composites, unconven-
tional techniques like infiltration of pre-forms by siliconorganic polymers are
required to produce composites and their components. Related to this an
important advantage of the polymer route in ceramics technology is that
forming processes can be outlined with melts of the pre-ceramic polymer at
very low temperatures in the range 150-250 °C; for instance melt-spinning of
fibers, liquid infiltration process for composite matrix production and spin-
coating of thin layers. That means complex parts - also with extreme
dimensions like fibers - can be made with nearly final size, and expensive
finishing can be considerably reduced.
The scheme in Fig. 11 shows several ways in which SiC components and
structures can be made. (Technologically preferred routes are represented by
thicker lines.)
In previous sections several Si-organic precursors for SiC fibers have been
compared. In Fig. 12 the processing steps of SiC-fiber production starting
processing of SiC-componenls
S i - 11 i l l l!. i ::• i-ir

powder route procure CVTOCVI
OF ( - | > r t f i M - n i -
1
SiC-powder + •siliconorganic • reaction of Si- and • infillraliori of carbonous
sinlering aids + precursor + C'Con[aining organics prc-form by liquidus Si
binder + (filler +) • ••' high temperatures or Si-vapour +
sinlering/ hol-pressing pyrolysis reaction
Fig. 11. Processing routes for SiC components and structures
from polymeric precursor are schematically drawn, illustrated by the route

applied at Freiberg University.
The special "Freiberg way" of fiber production - starting from disilane
fraction of MULLER-ROCHOW-synthesis - is a very promising access for
cheap SiC-fiber processing [249]. After melt spinning the poly(chlorometh-
ylsilane)-precursor green fiber must be cured to avoid melting during
pyrolysis. This step is critical for the whole process. Curing can be performed
by irradiation (electron beam) or aided by several reactive agents like oxygen
or ammonia, depending on the purpose of use. The pyrolysis is outlined under
inert atmosphere up to temperatures about 1000 °C to transform the
Si-organic network into an amorphous or nanocrystalline silicon carbide.
A further technological advantage of applying the polymer route for
composite matrices is that no mechanical damage of the reinforcing fibers is
caused. The scheme in Fig. 13 demonstrates the steps for making composites
by the liquid polymer infiltration process [250, 251].
The liquid infiltration process (LPI) has to be repeated sometimes due to
mass loss and volume shrinkage of the polymer during pyrolysis. Often the
liquid polymer is mixed with fillers (passive or active) to inhibit or to
compensate shrinkage [252].
The chemical vapor infiltration (CVI) of porous pre-forms - also a low
temperature/low pressure route - is also applied to make composite matrices
or in combination with LPI to raise the density of the pre-infiltrated structures.
The mechanical properties achieved by CVI are better compared with the ones
obtained by LPI, but CVI takes long processing time and is therefore not
usable for big plane and complex structures [253, 254].
scheme of processes of SiC fiber manufacture
disilane
fraction
from Miiller-
Rochow-
Synthesis
synthesis of
polysilane
spinning of
green
fibers
curing
pyrolysis
under
argon
Fig. XI. SiC-fiber processing from Si-organic precursor: "Freiberg way". (Courtesy of
H.-P. Martin, TU Freiberg, Germany)
The infiltration of carbonaceous structures by liquid silicon or silicon vapor

is also an interesting technique to produce SiC components. Especially the
Si-infiltration of C-structures coming from natural materials like wood offers
the possibility for low cost fabrication of SiC, and has already been described
in Sect. 4. Infiltration of polymer-matrix derived carbon fiber reinforced
carbon is schematically drawn in Fig. 14 [255, 256]. The production of
complex SiC-composite parts and components as well as big parts is possible
by this route.
Short fiber-, whisker-, platelet-, and particle-reinforced brittle matrix
composites are mainly hot-pressed.
Metal matrix composites with SiC-reinforcements are obtained by
melt-infiltration or powder metallurgical routes like hot-isostatic pressing.
SiC-fibers in metal matrix composites are often coated to prevent reaction
Si-precursor
solvent
1
shiny structures by
filament winding
n L.iiiiii powder
Impromitkin
Filament winding -
lamination of
fibers fiber coating [ ' prepress
curing of the structure

pressurtless pyrotysis
under autoclave oxidation 111-=-. •- •
conditions without forming took structure

protection
iOO-300'C from CMC
111.:II bur 1100-I4OTC
Fig. 13. Liquid polymer infiltration route for fabrication of SiC composite structures from
Si-precursors [from 251]
with molten metals and diffusion. MMCs will be left out of our
consideration in the following; for further information see [257, 258]
5.3
Properties of SiC Fibers and their Coatings
Since Yajima's [259] development of small diameter long SiC-based fibers

made by melt-spinning polysilanes following curing by oxygen and pyrolysis, a
lot of investigations have been made in that field of deriving ceramic fibers
from siliconorganic precursors. In the last few years the special goal was the
stabilization of the microstructure of the ceramic fiber at temperatures above
1200 °C. Relation between tailoring microstructure and tailoring high-tem-
perature properties of SiC-fibers has been already discussed in more detail in
Sect. 3. The so-called small diameter SiC fibers obtained by the polymer route
have diameters in the range from 10 to 15 um, and their mechanical properties
at room temperature allow winding procedures and fabrication of fabrics.
Figure 15 shows a SiC-fiber made by the "Freiberg way".
A comparison between SiC-fibers on market and in development is given
in Table 4 and should summarize the international activities in this field
[260-263].
Since the long term durability of CMCs is dominated by strength and creep
resistance of reinforcing fibers, fibers are exposed to several tests to evaluate
liquid precursor fiber pre-form
resin injection/
autoclave process
200'C S-20bar
pressureless pyrolysis
900*C nitrogen
c * rut 4'.
coupled with
chemical reaction
1550'C vacuum
integral structure
from CMC
Fig. 14. Fabrication of SiC components through Si-infiltration of C-pre-forms [from 255]
the behavior under high temperature conditions. Significant features obtained

from such tests are:
1. Relative mass loss vs temperature (thermal analysis) and composition
change (phase analysis)
2. Crystallite size vs T
3. Change of mechanical properties vs T (drop of strength and stiffness)
4. Creep behavior (elongation vs time, load) vs T
5. Oxidation behavior (mass gain, SiO- layer thickness growth) vs T
Properties of SiC-fibers listed in Table 4 are given in Table 5. Raised oxygen
content of SiC-fibers like NICALON and TYRANNO - introduced for curing -
limits the application temperature to about 1200 °C because of a significant
drop of mechanical properties. While oxygen-rich SiC fibers degrade due to
decomposition of the oxygen containing intergranular Si-C-O-phase, and
related to this undergo crucial changes of microstructure (grains grow,
porosity is developed, phases lose connection), leading in the end to fiber
destruction, electron-beam cured SiC-fibers have small oxygen contents and
are more stable, but only in non-oxidative environments due to their excess
carbon content. Outgasing oxygen mainly as CO during sintering at
120 G. Roewer • U. Herzog • K. Trommer • E. Mfiller • S. Fruhauf
Fig. 15. SiC-fiber made by the "Freiberg way" (SEM)
temperatures above 1700 °C is another way to lower the oxygen content in

oxygen cured SiC fibers. The high temperature properties of the obtained
fibers are excellent; see also Table 5.
For SiC-fibers in ceramic matrices - especially SiC - pyrolytic carbon and
BN (BN also for glass and glass-ceramic matrices) have been the only
successful interphase materials realizing the deflection of matrix cracks in the
interphase layer [264, 265]. Currently, substitution of pyrolytic carbon by BN
is being researched rather intensively, because oxidation of the carbon coating
in the temperature region 700-800 °C allowed by matrix cracks results in
embrittlement of the CMC [266]. However, BN-coatings also degrade at
temperatures below 1000 °C due to high oxygen contents coupled with
extensive silica formation at the interface followed by embrittlement of
the composite [267]. A way to protect C- and BN-layers from oxidative
degradation is to deposit SiC on the first functional fiber coating and to
construct multi-layer coatings [268-270].
5.4
Properties and Applications of SiC Reinforced Glass and Glass-Ceramics
The reinforcement of glasses and glass-ceramics with SiC-fibers has been

investigated for about 20 years. Table 6 contains the mechanical properties of
SiC-reinforced Duran-glass, BMAS, and LAS-glass-ceramics [271-276].
The fabrication of glass or glass ceramic matrices by hot pressing is outlined
at moderate temperatures and pressures, but they are sufficiently high to cause
formation of interphases between SiC-fibers - namely NICALON and
TYRANNO - their coatings and the matrix. Formation of very complex
CO CO
0.2
I 1
o
i =3
V v o CO*
in
m
i
CN
i
I
•a
2.5-
1
0.5-
o o
i-H CN i—( O
T^ CO
i-H (N --H O
00 CO
f^ ^ CO
'O CO O
o
•z,
< w
£2 uS K
LO CO
=J =>
iTi fN
60
u.S
II ^ q ^
£ I O CN *-H CO
o
I-H O\
2;
o
o
6
c/5
CO LO
u
B | *
•a -ts a
a
Table 5. Properties of small-diameter SiC-fibers
NICALON Hi-NICALON Hi-NICALON SYLRAMIC TYRANNO AM SA ZMI ZE UF UF-HM

CG type S Dow corning Lox E
Rt-modulus (GPa) 220 273 390 386 199 180 420 200 233
Rt-strength (GPa) 3.0 2.8 2.6 3.4 2.9 2.8 2.8 3.4 3.5 3.23 3.0
Elongation (%) 1.4 1.0 0.6 0.75 1.5 1.6 0.7 1.7 1.5
Ht-strength (GPa) 2.94 1.75 1.75
Electrical resistance Clem 103-104 1.4 0.1 0.8 2.0
Lin. thermal expansion K 3.2 3.5 5.4
Heat conductivity 2.97 7.77 18.4 40-45 2.4 64.6 2.52
(20 °C)
2.20 10.1 16.3
E
Table 6. Mechanical properties of SiC-reinforced glasses and glass-ceramics
Property SiC/DURAN® SiC/BMAS SiC/LAS
Volume fraction of fiber ( %) 40 35-45 35-45

Density (g cm"3) 2.5
Elongation (%) 0.5-1 0.37 0.26
Tensile modulus (gpa) 110 149 125
Flexure strength (mpa) 450 221 316
Interfacial shear strength (mpa) 106 119
Linear thermal expansion coefficient (10~6 K"1): 3.5
Heat conductivity (Wm^K" 1 ) 1.5-3
BMAS - barium-magnesium-aluminosilicate glass ceramic

LAS - lithium-aluminosilicate glass ceramic)
Fig. 16. Si-terminated 6H-SiC and DURAN glass (lattice plane resolved micrograph).
(Courtesy of J. Woltersdorf, Max-Planck-Institut, Halle)
interphase structures consisting of C and SiO2 due to degradation of Si-C-O-

grain boundary phase of the oxygen-rich SiC-fiber in glass matrices and
oxidation of SiC above 1000 °C are described by many authors [273, 274].
These phenomena have also been thoroughly investigated by Woltersdorf et al.
[275]. For the oxidation process of SiC, a four-step model has been developed
and verified for an idealized system (mono-crystalline 6H-SiC/Duran glass) by
methods of analytical TEM. Figure 16 gives a lattice-plane resolved micro-
graph of the interface of Si-terminated 6H-SiC [001] and Duran glass revealing
the existence of graphite-like carbon within the glass matrix separated by a
silica-rich layer from SiC.
The maximum temperature of use for NICALON- or TYRANNO-fiber
reinforced glasses and glass-ceramic is limited by the transformation
temperature of the glass phases, i.e., about 500 °C for SiC/borosilicate-glasses,
Table 7. Mechanical properties of C/C-SiC, SiC(LPI), and SiC (CVI) composite materials
Property C/SiC (LPI) C/SiC (CVI) C/C-SiC

(Si-infiltration)
Open porosity (%) 10-30 10-15 2-5

Volume fraction of fiber ( %) 30-60 42-47 55-65
Density (g arT 3 ) 1.9-2.0 2.1-2.2 1.8-1.95
Tensile strength (mpa) 230-260 270-330 120-190
Elongation (%) 1-2.5 0.6-0.9 0.2-0.35
Tensile modulus (gpa) 80-90 90-100 50-70
Flexure strength (mpa) 500-600 450-500 200-290
Interlaminar shear strength (mpa) 15-30 45-55 25-30
Linear thermal expansion
coefficient (10~6 K"1)
parallel to fiber axis 1.7 3 1-2
perpendicular to fiber axis 5.6 5 4-6
Heat conductivity (WnT1 K"1)
parallel to fiber axis 10-15
perpendicular to fiber axis 6-8
LPI-liquid polymer infiltration

CVI-chemical vapor infiltration
Fig. 17. Integrally laminated edge-panel of a spacecraft heat shield, edge length 400 mm.
(Courtesy of A. Miihlratzer, MAN Technology, Germany)
700 °C for aluminosilicate-glasses, and 900 °C for glass-ceramics. The C-rich

interphase supports crack deflection at the interface, but oxidation also
happens here which leads to limited lifetime of the composite. Good wear
properties distinguish SiC-fiber reinforced glasses like SiC/Duran so they can
be used, for example, as a substitute for asbestos in glass fabrication, for
gripping and sliding devices manipulating hot glass parts [276].
5.5
Properties and Application of C/SiC and SiC/SiC Composites
Table 7 contains a comparison of mechanical properties achieved for the

three unconventional routes described to make SiC, composite matrices by
LPI (liquid polymer infiltration), CVI (chemical vapor infiltration) and
Si-infiltration [277].
From the present status of technology, the components and structures with
long C- and SiC-fiber reinforced SiC-matrices have sufficient fhermomechan-
ical properties, especially high fracture toughness for thermally stressed
structures under complex demands and temperatures above 1200 °C. Thus,
bearings, gas turbine parts, or heat-shield-components can be made from
C/SiC and SiC/SiC composite materials. The edge panel of a heat shield, shown
in Fig. 17, illustrates the possibilities of the infiltration techniques - here the
LPI - to make very complex and big architectures [254].
Figure 18 offers an interesting insight into the microstructure of the edge
panel shown above. The good macroscopic thermomechanical properties of
that part are based on a very inhomogeneous microstructure, which shows
a porosity in the range of 10-15% even after reinfiltration [253].
A special field of application for SiC-based CMC is, beside aerospace and
aviation, also the automotive industry. The substitution of cast iron brake
discs by C/C-SiC brake discs became attractive, for example, for high-speed
trains due to savings of weight, appropriate friction properties, and better
high-temperature stability. C/SiC composite brake discs are made by CVI, LPI,
or mainly by Si-infiltration of carbon-fiber reinforced carbon matrix com-
posites as described in Sect. 5.2 [253, 278, 279]. The application of SiC-based
ceramic matrix composites to an automotive ceramic gas turbine has been
thoroughly discussed by Kaya [280]. SiC-composites are applied to turbine
rotor, back plate, orifice and extension liner, and for inner scroll support.
Investigations of thermomechanical properties of these components have
confirmed that CMC is superior in various characteristics to a monolithic
ceramic. They show highly reliable resistance to damage and to thermal shock.
Also the special possibilities in fabrication like controlling the orientation of
reinforcing fibers or whiskers could be used for this purpose.
5.6
Porous SiC for Membranes and Filters
Porous SiC structures are natural products resulting from pyrolysis of

polysilanes because of high mass loss through outgasing, increasing density,
and volume shrinkage simultaneously to the forming of an unmeltable
polycarbosilane and amorphous/crystalline SiC. The pore size resulting from
outgasing can be controlled by polymer composition (for example by
variation of side- and end-groups) and pyrolysis conditions. SiC membranes
and filters are intended for gas separation, cleaning-up dirty gases, and
removing particles from gases, liquids, or melts, depending on average pore
size and volume, pore structure, and pore surface properties. Both SiC
Fig. 18. Microstructure of a two-dimensional C-fiber reinforced SiC-composite (150x).

(Courtesy of A. Muhlratzer, MAN Technology, Germany)
Fig. 19. Microstructure of a SiC foam ceramic for filters and other applications [283]
membranes with smaller pore size and SiC filters with higher pore size possess
the advantage to operate under high-temperature conditions and to allow a
high number of regeneration cycles. While SiC-membrane fabrication via the
polymer route has just begun, several routes for filter fabrication are known.
For example so-called foam-ceramic organic (polymeric or natural) pre-
structures are pyrolyzed to carbon, and then they are infiltrated either by
liquid silicon or silicon vapor together with simultaneous reaction of Si and C
to SiC. Using polysilanes, the pyrolysis can be outlined in special cases after
infiltration to summarize the transformation organic-inorganic in one
processing step. Fiber reinforced SiC-matrix composites are modified for
filter applications stopping the densification during the infiltration process of
the pre-form [281, 282].
The microstructure of an SiC-filter made from silicon infiltrated polymer
foam is shown in Fig. 19. Cell size, cell geometry, and cell anisotropy is
controllable during processing [283]. The structural variability of this material
reaches from tube-like anisotropic to isotropic pore nets with several pore and
bridge averages. This porous SiC material cannot only be applied to niters and
membrane supports; it can also be used as catalyst support or heat exchanger
[284].
6
References
1. Yajima S, Hayashi J, Omori M, Okamura K (1976) Nature 261: 683
2. Laine RM, Babonneau F (1993) Chem Mater 5: 265
3. a) Giorgis F, Giuliani F, Pirri CF, Tresso EJ, Conde JP, Chu V (1998) J Non-Cryst Solids
465: 227-230; b) Kanicki J (ed) (1991) Amorphous and macrocrystalline semiconductor
devices, Vol 1 Artech House, Boston, MA
4. Pasche S, Toyama T, Okamoto H, Hamakawa Y (1989) IEEE Trans Electron Dev 36: 895
5. Solomon I, Schmidt MP, Tran-Quoc H (1988) Phys Rev B38: 9895
6. Nakaaki I, Saito N, Inui Y, Yoshioka S, Nakamura S (1993) Philos Mag B68: 55
7. Chu V, Conde JP (1996) Amorphous silicon technology. Materials Research Society
Symposium Proc 420: 381, Pittsburgh, PA 1996
8. Kumbhar AA, Dusane RO, Bauer S, Schroder B (1998) J Non-Cryst Solids 452: 227-230
9. Schmidbaur H, Hager R, Zech J (1991) In: Bassindale AR, Gaspar PP (eds) Frontiers in
Organosilicon Chemistry 62. Royal Society of Chemistry, Cambridge
10. Bastl Z, Burger H, Fajgar R, Pokorna D, Pola J, Senzlober M, Subrt J, Urbanova M
(1996) Appl Organomet Chem 10: 83
11. Larkin DJ, Interrante LV (1992) Chem Mater 4: 22
12. Boo JH, Kim KSY, Yenon SH, Jung NI (1995) Chem Mater 7: 694
13. Volnina EA, Kupcik J, Bastl Z, Subrt J, Gusel'nikov L, Pola J (1997) J Mater Chem 7: 637
14. Scholz VM, Fuss W, Kompa KL (1993) Adv Mater 5: 38
15. Pola J, Vitek J, Bastl Z, Urbanova M, Subrt J, Taylor R (1997) J Mater Chem 7: 1415
16. Drinek V, Bastl Z, Subrt J, Pola J (1998) Appl Organomet Chem 12: 427
17. Hattori Y, Kruangam D, Tayama T, Okamoto H, Hamakama Y (1987) J Non-Cryst
Solids 97/98: 1079
18. Hattori Y, Kruangam D, Kato K, Nitta Y, Okamoto H, Hamakawa Y (1987) Proc 19th
IEEE Photovoltaic Specialists Conf, New Orleans 689
19. Kimoto T, Nishino H, Yamashita A, Woo WS, Matsunami H (1989) In: Yang CY,
Rahman MM, Harris GL (eds) Springer Proceedings in Physics 71, Proc Amorphous
and Crystalline Silicon Carbide IV: 31. Springer, Berlin, Heidelberg, New York
20. Rowland LB, Tanaka S, Kern RS, Davis RF (1992) Springer Proceedings in Physics 71,
Proc Amorphous and Crystalline Silicon Carbide IV:84, Santa Clara, CA. Springer,
Berlin, Heidelberg, New York
21. Tran A (1989) In: Harris GL, Yang CYW (eds) Springer Proceedings in Physics 34, Proc
Amorphous and Crystalline Silicon Carbide 1:167. Springer, Berlin, Heidelberg, New York
22. Manning BM, Hewitt SB, Tarr NG, McElwee TW (1989) In: Yang CY, Rahman MM,
Harris GL (eds) Springer Proceedings in Physics 71, Proc Amorphoous and Crystalline
Silicon Carbide IV 232. Springer, Berlin Heidelberg New York
23. a) Gandia JJ, Gutierrez, Carabe J (1993) Solar Energ Mater Solar Cells 29: 139;
b) Nakayama Y, Yajima T, Ueda H, Akita S, Kawamura T (1991) J Non-Cryst Solids
1279:137/138; c) Tews R, Suchanek G, Kottwitz A, Abramov AS, Kosarev AI (1998)
J Non-Cryst Solids 478: 227-230
24. a) Giorgis F, Giuliani F, Pirri CF, Tresso E, Conde JP (1998) J Non-Cryst Solids 465:
227-230; b) Senemaud C, Georghiu de la Rocque A, Bonnefont (1998) J Non-Cryst
Solids 447: 227-230
25. Scarlete M, Butler IS, Harrod JF (1995) Chem Mater 7: 1214
26. Furukawa S, Obana M, Nakamine T, Shirakawa Y, Sorai A, Tamura M (1992) J Phys
Condens Matter 4: 5167
27. a) Watanabe A, Matsuda M (1991) Chem Lett 1101; b) Kani R, NakanoY, Hayase S
(1995) Macromolecules 28: 1773
28. Colombo P, Paulson TE, Pantano CG (1997) J Am Ceram Soc 80: 2333
29. Pivin JC, Colombo P, Tonidandel M (1996) J Am Ceram Soc 79: 1967
30. West R (1986) J Organomet Chem 300: 327
31. a) Zeigler JM, Harrah LA, Johnson AW (1987) Polymer Preprints 28: 424; b) Zeigler JM,
McLaughlin LI, Perry RJ (1991) J Inorg Organomet Polym 1: 531; c) Jones RG,
Benfield RE, Cragg RH, Swain AC, Webb SJ (1993) Macromolecules 26: 4878
32. Jones RG, Wong WKC (1998) Organometallics 17: 59
33. a) Miller RD, Ginsburg EJ, Jenkner P, Thompson D (1993) Polymer Preprints 34: 232;
b) Miller RD, Thompson D, Sooriyakumaran R, Fickes GN (1991) J Polym Sci A 29: 813
34. a) Cragg RH, Jones RG, Swain AC, Webb SJ (1990) J Chem Soc Chem Commun 1147;
b) Miller RD, Ginsburg EJ, Thompson D (1993) Polym J 25: 807; c) Gauthier S,
Worsfold DJ (1989) Macromolecules 22: 2213
35. a) Kim HK, Matyjaszewski K (1988) J Am Chem Soc 110: 3321; b) Matyjaszewski K,
Greszta D, Hrkach JS, Kim HK (1995) Macromolecules 28: 59
36. Kipping FS (1924) J Chem Soc 125: 2291
37. Burkhard CA (1949) J Am Chem Soc 71: 963
38. a) Yajima S, Hayashi J, Omori M (1975) Chem Lett 931; b) Yajima S, Okamura K,
Hayashi J (1975) Chem Lett 1209
39. Yajima S (1983) Am Ceram Soc Ceram Bull 62: 893
40. Yajima S, Hayashi J, Okamura K (1977) Nature 266: 521
41. Ichikawa H, Machino F, Mitsuno S, Ishikawa T, Okamura K, Hasegawa Y (1986) J Mat
Sci 21: 4352
42. a) Sakurai H, Koh R, Hosomi A, Kumada M (1966) Bull Chem Soc Jpn 39: 2050;
b) Sakurai H, Hosomi A, Kumada M (1968) Chem Commun 930
43. Yajima S, Hasegawa Y, Hayashi J, Iimura M (1978) J Mater Sci 13: 2569
44. Okamura K (1987) Composites 18: 107
45. Soraru GD, Babonneau F, Mackenzie JD (1990) J Mater Sci 25: 3886
46. West R, David LD, Djurovich PI, Yu H (1983) Am Ceram Soc Ceram Bull 62: 899
47. Shieh Y-T, Sawan SP (1995) J Appl Polm Sci 58: 2013
48. West R, Nozue I, Zhang X-H, Trefonas P (1984) Polymer Preprints 25: 4
49. Schilling LC Jr (1986) Brit Polym J 18: 355
50. a) Seyferth D, Sobon CA, Borm J (1989) New J Chem 14: 545; b) Seyferth D, Yu J-F
(1987) US Pat 4,639,501
51. a) Bianconi PA, Weidman TW (1988) J Am Chem Soc 110: 2342; b) Bianconi PA,
Weidman TW (1989) Macromolecules 22: 1697
52. Matyjaszewski K, Chrusciel J, Maxka J, Sasaki M (1995) J Inorg Organomet Polym 5: 261
53. Maxka J, Chrusciel J, Sasaki M, Matyjaszewski K (1994) Macromol Symp 77: 79
54. Trandell RF, Urry G (1978) J Inorg Nucl Chem 40: 1305
55. Baney RH, Gaul JH, Hilty TK (1983) Organometallics 2: 859
56. a) Herzog U, Richter R, Brendler E, Roewer G (1996) J Organomet Chem 507: 221;
b) Richter R, Roewer G, Bohme U, Busch K, Babonneau F, Martin H-P, Muller E (1997)
Appl Organomet Chem 11: 71
57. Richter R, Schulze N, Roewer G, Albrecht J (1997) J Prakt Chem 339: 145
58. Herzog U, Schulze N, Trommer K, Roewer G (1999) Main Group Metal Chemistry 22: 19
59. a) Hengge E, Kalchauer W, Schrank F (1986) Monatsh Chem 117: 1399; b) Herzog U,
Schulze N, Trommer K, Roewer G (1997) J Organomet Chem 547: 133
60. a) Kabeta K, Wakamatsu S, Imai T (1997) J Polym Sci A, Polym Chem 35: 455;
b) Kobayashi T, Hatayama K, Suzuki S, Abe M, Watanabe H, Kijima M, Shirakawa H
(1998) Organometallics 17: 1646
61. a) Atwell WH, Weyenberg DR (1968) J Am Chem Soc 90: 3438; b) Watanabe H, Inose J,
Muraoka T, Saito M, Nagai Y (1983) J Organomet Chem 244: 329
62. a) Watanabe H, Higuchi K, Goto T, Muraoka T, Inose J, Kageyama M, Iizuka Y, Nozaki
M, Nagai Y (1981) J Organomet Chem 218: 27; b) Watanabe H, Abe M, Sonoda K,
Uchida M, Ishikawa Y, Inomiya M (1991) J Mater Chem 1: 483
63. Kalchauer W, Pachaly B, Geisberger G, Rosch L (1992) Z Anorg Allg Chem 618: 148
64. a) Marciniec B (ed) (1992) Comprehensive handbook on hydrosilylation. Pergamon,
Oxford; b) Ojima I (1989) In: Patai S, Rappoport Z (eds) The Chemistry of Organic
Silicon Compounds, Wiley, Chinchester, pl479; c) Laine RM, Babonneau F (1993)
Chem Mater 5: 260
65. a) Lickiss PD (1992) Chem Soc Rev 271; b) Tobita H, Wada H, Ueno K, Ogino H (1994)
Organometallics 13: 2545; c) Pannell KH, Brun MC, Sharma H, Jones K, Sharma S
(1994) Organometallics 13: 1075; d) Tamao K, Tarao Y, Nakagawa Y, Nagata K, Ito Y
(1993) Organometallics 12: 1113
66. a) Kiso Y, Tamao K, Kumada M (1974) J Organomet Chem 76: 95; b) Kobayashi TA,
Yamashita H, Tanaka M (1988) Chem Lett 1411; c) Ojima I, Inaba SI, Kogure T, Nagai Y
(1973) J Organomet Chem 55: C7; d) Brown-Wensley KA (1987) Organometallics
61: 590
67. a) Yamashita H, Kobayashi T, Hayashi T, Tanaka M (1990) Chem Lett 1447; b) Pan Y,
Mague JT, Fink MJ (1992) Organometallics 11: 3495; c) Murakami M, Yoshida T,
Ito Y(1994) Organometallics 13: 2900; d) Ozawa F, Sugawara M, Hayashi T (1994)
Organometallics 13: 3237; e) Kobayashi TA, Hayashi T, Yamashita H, Tanaka M (1989)
Chem Lett 467; f) Kiso Y, Tamao K, Kumada M (1974) J Organomet Chem 76: 105
68. Gauvin F, Harrod JF, Woo HG (1998) Adv Organomet Chem 42: 363
69. Sakakura T, Lautenschlager HJ, Nakajima M, Tanaka M (1991) Chem Lett 913
70. Kobayashi T, Sakakura T, Hayashi T, Yumura M, Tanaka M (1992) Chem Lett 1157
71. Nakano T, Nakamura H, Nagai Y (1989) Chem Lett 83
72. Bourg S, Corriu RJP, Enders M, Moreu JJE (1995) Organometallics 14: 564
73. a) Aitken C, Harrod JF, Gill US (1987) Can J Chem 65: 1804; b) Dioumaev VK,
Harrod JFJ (1996) J Organomet Chem 521: 133
74. a) Aitken CT, Harrod JF, Samuel E (1986) J Am Chem Soc 108: 4059; b) Harrod JF,
Yun SS (1987) Organometallics 6: 138; c) Aitken CT, Harrod JF, Samuel EJ (1985)
J Organomet Chem 279: Cll; d) Aitken CT, Harrod JF, Samuel EJ (1986) Can J Chem
64: 1677
75. Zhang ZF, Babonneau F, Laine RM, Mu Y, Harrod JF, Rahn JA (1991) J Am Ceram Soc
74: 670
76. a) Seyferth D, Wood TG, Tracy HJ, Robinson L (1992) J Am Chem Soc 75: 1300;
b) Seyferth D, Sobon CA, Borm J (1990) New J Chem 14: 545
77. Gauvin F, Harrod JF, Woo HG (1998) Adv Organomet Chem 42: 363
78. a) Tilley TD (1993) Ace Chem Res 26: 22; b) Tilley TD, Hoo HG, (1989) J Am Chem Soc
111: 3757; c) Tilley TD, Hoo HG (1989) J Am Chem Soc 111: 8043; d) Walzer JF, Tilley
DT, Hoo HG (1992) J Am Chem Soc 114: 7047; e) Imori T, Tilley TD (1994) Polyhedron
13: 2231
79. a) Aitken C, Harrod JF, Gill US (1987) Can J Chem 65: 1804; b) Dioumaev VK, Harrod
JF (1996) J Organomet Chem 521: 133
80. a) Corriu RJP, Enders M, Huille S, Moreau JJE (1994) Chem Mater 6: 15; b) Campbell
WH, Hilty TK, Urga L (1989) Organometallics 8: 2615
81. Gauvin F, Harrod JF (1990) Can J Chem 68: 1638
82. a) Li H, Gauvin F, Harrod JF (1993) Organometallics 12: 575; b) Corey JY, Zhu X,
Bedard TC, Lange LD, (1991) Organometallics 10: 924; c) Corey JY, Zhu X (1992)
Organometallics 11: 672
83. a) Woo HG, Song SJ (1996) J Korean Chem Soc 17: 494; b) Woo HG, Song SJ (1997)
J Korean Chem Soc 41(9): 502; c) Ojima I, Inaba SI, Kogure T, Nagai Y (1973)
J Organomet Chem 55: 55
84. a) CyprykM, Gupta Y, Matyjaszewski K (1991) J Am Chem Soc 113: 1046; b) Chrusciel J,
Fossum E, Matyjaszewski K (1993) Polymer Preprints 34: 221; c) Fossum E, Matyjaszew-
ski K (1995) Macromolecules 28: 1618; d) Matyjaszewski K, Cypryk M, Frey H, Hrkach J,
Kim HK, Moeller M, Ruehl K, White M (1991) J Macromol Sci-Chem A28: 1151
85. Kunai A, Kawakami T, Toyoda E, Ishikawa M (1991) Organometallics 10: 2001
86. Kunai A, Kawakami T, Toyoda E, Ishikawa M (1991) Organometallics 10: 893
87. Biran C, Bordeau M, Pons P, Leger M-P, Dunogues J (1990) J Organomet Chem 382: C17
88. Shono T, Kashimura S, Ishifune M, Nishida R (1990) J Chem Soc, Chem Commun 1161
89. Jammegg C, Graschy S, Hengge E (1994) Organometallics 13: 2397
90. Kimata Y, Suzuki H, Satoh S, Kuriyama A (1995) Organometallics 14: 2506
91. a) Sakamoto K, Obata K, Hirata H, Nakajama M, Sakurai H (1989) J Am Chem Soc
111: 7641; b) Sakurai H, Sakamoto K, Funada Y, Yoshida M (1993) Polymer Preprints
34: 218; c) Sakurai H, Yoshida M, Sakamoto K (1996) J Organomet Chem 521: 287
92. Roark DN, Peddle GJD (1972) J Am Chem Soc 94: 5837
93. van Aefferden B, Habel W, Sartori P (1990) Chemiker Ztg 114: 367
94. a) Habel W, Mayer L, Sartori P (1991) Chemiker Ztg 115: 301; b) Habel W, Haeusler W,
Sartori P (1995) J Organomet Chem 492: 225
95. Habel W, Oelschlager A, Sartori P (1995) J Organomet Chem 486: 267
96. Habel W, Mayer L, Sartori P (1994) J Organomet Chem 474: 63
97. Habel W, Harnack B, Nover C, Sartori P (1994) J Organomet Chem 467: 63
98. a) Hemida AT, Birot M, Pillot JP, Dunogues J, Pailler R, Naslain R (1997) J Mater Sci
32: 3485; b) Hemida AT, Birot M, Pillot JP, Dunogues J, Pailler R (1997) J Mater Sci
32: 3475
99. a) Ijadi-Maghsoodi S, Pang Y, Barton TJ (1990) J Polym Sci A - Polym Chem 28: 955;
b) Habel W, Moll A, Sartori P (1997) J Prakt Chem 339: 291
100. Ijadi-Maghsoodi S, Barton TJ (1991) Macromolecules 23: 4485
101. Tretner C, Zobel B, Hummeltenberg R, Uhlig W (1994) 468: 63
102. Uhlig W (1997) J Organomet Chem 545: 281
103. Oshita J, Yamashita A, Hiraoka T, Shinpo A, Kunai A, Ishikawa M (1997)
Macromolecules 30: 1540
104. Uhlig W (1999) J Organomet Chem 130: 181
105. Liao CX, Chen MW, Weber WP (1993) Polymer Preprints 34: 230
106. Myers AG, Kephart SE, Chen H (1992) J Am Chem Soc 114: 7922
107. a) Vdovin VM, Pushchevaya KS, Belikova NA, Sultanov R, Plate AF, Petrov AD (1961)
Dokl Akad Nauk SSSR 136: 96; b) Nametin NS, Vdovin VM, Zav'yalov VI (1965) Dokl
Akad Nauk SSSR 162: 824
108. a) Nametin NS, Vdovin VM, Zav'yalov VI (1964) Izv Akad Nauk SSSR Ser Khim 203; b)
Nametin NS, Bespalova NB, Ushakov NV, Vdovin VM (1973) Dokl Akad Nauk SSSR
209: 621; c) Weber WP (1992) Polym Bull 28: 281
109. a) Nametin NS, Vdovin VM, Grinberg PL (1964) Izv Akad Nauk SSSR Ser Khim 1133;
b) Weyenberg DR, Nelson LE (1965) J Org Chem 30: 2618; c) Cundy CS, Eaborn C,
Lappert MF (1972) J Organomet Chem 44: 291; d) Poletaev VA, Vdovin VM, Nametin
NS (1973) Dokl Akad Nauk SSSR 208: 1112
110. Ogawa T, Murakami M (1996) Chem Mater 8: 1260
111. Yamashita H, Tanaka M, Honda K (1995) J Am Chem Soc 117: 8873
112. Matsumoto K, Shimazu H, Deguchi M, Yamaoka H (1997) J Polym Sci A Polym Chem
35: 3207
113. Nametkin NS, GusePnikov LE, Vdovin WM, Grinberg PL, Zav'yalov VI, Oppengeim VD
(1966) Dokl Akad Nauk SSSR 171: 630
114. Bordeau M, Biran C, Pons P, Leger MP, Dunogues J (1990) J Organomet Chem 382: C21
115. Kriner WA (1964) J Org Chem 29: 1601
116. Interrante LV, Rushkin I, Shen Q (1998) Appl Organomet Chem 12: 695
117. a) Wu HJ, Interrante LV (1989) Chem Mater 1: 564; b) Whitmarsh CW, Interrante LV
(1991) Organometallics 10: 1336; c) Whitmarsh CW, Interrante LV (1992) US Pat
5153295
118. a) Rushkin I, Shen Q, Lehman SE, Interrante LV (1997) Macromolecules 30: 3141;
b) Interrante LV, Whitmarsh CK, Sherwood W (1995) Ceram Trans 58: 111
119. a) Singler RE, Schneider NS, Hagnauer GL (1984) Polym Eng Sci 60/61: 1; b) Babonneau
F, Bois L, Interrante LV (1994) Chem Mater 6: 51; c) Yang CY, Marchetti P, Interrante
LV (1992) Polym Prepr 33: 208
120. Lienhard M, Rushkin I, Verdecia G, Wiegand C, Apple T, Interrante LV (1997) J Am
Chem Soc 119: 12,020
121. a) Corriu R, Buory B, Carpenter L (1989) Chem Abstr 111: 8038; b) Boury B, Carpenter
L, Corriu R (1990) Angew Chem Int Edit Engl 29: 785; c) Boury B, Corriu RJP, Douglas
WE (1991) Chem Mater 3: 487; d) Boury B, Carpenter L, Corriu R, Mutin H (1990) New
J Chem 14: 535; e) Boury B, Corriu R, Leclercq D, Mutin P, Planeix JM, Viuox A (1991)
Organometallics 10: 1457
122. Pang Y, Ijadi-Maghsoodi S, Barton TJ (1993) Macromolecules 26: 5671
123. a) Bacqe E, Pillot JP, Birot M, Dunogues J, Lapouyade P, Bouillon E, Pailler R (1991)
Chem Mater 3: 348; b) Yashita H, Tanaka M (1995) Bull Chem Soc Jpn 6: 403
124. Schmidt WR, Interrante LV, Doremus RH, Trout TK, Marchetti PS, Maciel GE (1991)
3:257
125. Shinina K, Kumada M (1958) 23: 139
126. Carlson DJ, Cooney JD, Gauthier S, Worsfold (1990) J Am. Ceram Soc 73: 237
127. Togel D, Antony A, Bill J, Scheer P, Eichhofer A, Fritz G (1995) J Oganomet Chem
521: 125
128. Corriu RJP, Leclercq D, Mutin, Vioux A (1992) Chem Mater 4: 711
129. a) Yang LY, Interrante LV (1992) Polym. Prep. 33: 152; b) Interrante LV, Whitmarsh
CK, Trout TK, Schmidt R (1992) In: Laine RM (ed) Inorganic and Organometallic
Polymers with Spectroscopic Properties. NATO ASI Ser 206 (E). Kluwer, Academic,
London 1992: 243
130. Ting S-J, Chu C-J, Mackenzie JD (1992) J Mater Res 7: 164
131. a) Burns G, Chandra G (1989) J Amer Ceram Soc 72: 333; b) Taki T, Inui M, Okamura
K, Sato M (1989) J Mater Sci Lett 8: 1119
132. a) Babonneau F, Maquet J, Bonhomme C, Richter R, Roewer G, Bahloul D (1996) Chem
Mater 8: 1425; b) Matthews S, Edirisinghe MJ, Folkes MJ (1999) Ceramics Int 25: 49
133. Soraru GD, Babonneau F, Mackenzie JD (1990) J Mater Sci 25: 3886
134. Ting SJ, Chu CJ, Mackenzie JD (1992) J Mater Res 7: 164
135. Martin HP, Muller E, Richter R, Roewer G, Brendler E (1997) J Mater Sci 32: 1381
136. Chawla KK (1993) Ceramic Matrix Composites. Chapman and Hall, London, p 13
137. Vogel W (1992) Glaschemie. Springer, Berlin, Heidelberg, New York
138. Muller E, Heide K, Zanotto ED (1993) J Non-Cryst Solids 155: 56
139. a) Laine RM, Babonneau F (1993) Chem Mater 5: 260; b) Babonneau F (1995) In:
Harrod JF, Laine RM (eds) Applications of Organometallic Chemistry in the
Preparation and Processing of Advanced Material. NATO ASI series E, vol 297.
Kluwer, Dordrecht Boston London, p 103; c) Friess M, Bill J, Szabo DV, Riedel R (1994)
Key Eng Mater 89-91: 95; d) Seher M, Bill J, Aldinger F, Riedel R (1994) J Cryst Growth
137: 452; e) Xu Y, Zangvil A, Lipowitz J, Rabe A, Zank GA (1993) J Am Ceram Soc 76:
3034; f) Wang JW, Pirouz P (1993) J Mat Res 8: 2902
140. Martin HP (1993) Thesis, TU Freiberg
141. Martin HP, Muller E, Brendler E (1996) J Mater Sci 31: 4363
142. Martin HP, Irmer G, Muller E (1998) J Europ Ceram Soc 18: 193
143. Kurtenbach D, Martin HP, Muller E, Roewer G, Hoell A (1998) J Europ Ceram Soc
18: 1885
144. Kurtenbach D, Mitchell BS, Zhang H, Ade M, Muller E (1999) Thermochim Acta
337: 155
145. Mitchell BS, Zhang H, Malijkovic N, Ade M, Kurtenbach D, Muller E (1999) J Am
Ceram Soc 82: 2249
146. Kurtenbach D (2000) Thesis, TU Freiberg
147. Monthioux M, Delverdier O (1996) J Europ Ceram Soc 16: 721
148. Delverdier O, Monthioux M, Mocaer D, Pailler R (1993) J Europ Ceram Soc 12: 27
149. Izumi T, Show Y, Deguchi M, Kitabatake M, Hirao T, Mori Y, Ito T, Hiraki A (1993)
Thin Solid Films 228: 285
150. Soraru GD, Babonneau F, Mackenzie JD (1990) In: Zelinski BJJ et al. (eds) Better
ceramics through chemistry IV, MRS Symp Proceed, vol 180, p 815
151. Marotta A, Buri A, Branda F (1981) J Mater Sci 16: 341
152. Ray CS, Day DE (1990) J Am Ceram Soc 73: 439
153. Xu XJ, Ray CS, Day DE (1991) J Am Ceram Soc 74: 909
154. Kissinger HE (1956) J Res Nat Bur Stand (US) 57: 217
155. Hollemann-Wiberg (1985) Lehrbuch der Anorganischen Chemie, De Gruyter, Berlin,
p 139
156. Yoshi K (1991) Thin Solid Films 199: 85
157. Hon JD, Davis RF, Newburry DE (1981) J Mater Sci 16: 2485
158. Delvedier O, Monthioux M, Mocaer D, Pailler R (1994) J Europ Ceram Soc 14: 313
159. Kleebe HJ (1998) Phys Stat Sol (a) 166: 297
160. Kleebe HJ, Suttor D, Muller H, Ziegler G (1998) J Am Ceram Soc 81: 2971
161. Kleebe HJ, Suttor D, Ziegler G (1999) In: Bill J, Wakai F, Aldinger F (eds) Precursor-
Derived Ceramics. Wiley-VCH Weinheim, p 113
162. Bill J, Aldinger F (1999) In: Bill J, Wakai F, Aldinger F (eds) Precursor-Derived
Ceramics. Wiley-VCH Weinheim, p 33
163. Martin HP, Muller E, Dachselt U (1999) J Mater Sci 34: 2665
164. Martin HP, Muller E (1999) J Mater Sci 34: 2671
165. Babonneau F, Soraru GD, Thorne KJ, Mackenzie JD (1991) J Am Ceram Soc 74: 1725
166. Granozzi G, Glisenti A, Soraru GD (1990) In: Zelinski BJJ et al (eds) Better Ceramics
through Chemistry IV. MRS Symp Proceed, Vol 180, p 811
167. Berger MH, Hochet N, Bunsell AR (1999) In: Bunsell AR, Berger MH (eds) Fine ceramic
fibers. Marcel Dekker, New York Basel, p 231
168. Martin HP, Muller E, Irmer G, Babonneau F (1997) J Europ Ceram Soc 17: 659
169. Schwetz KA (2000) In: Riedel R (ed) Handbook of Ceramic Hard Materials, Vol 2.
Wiley-VCH, Weinheim, p 683
170. Weimer AW (1997) In: Weimer AW (ed) Carbide, Nitride and Boride Materials -
Synthesis and Processing. Chapman and Hall, London, p 94
171. Metselaar R, van Beck JA, Kodentsov A, van Loo FJJ (1994) Transact Mat Res Soc Japan
14A: 809
172. Prener JS (1963) US Pat 3 085 863
173. Kevorkian M, Komac M, Kolar D (1992) J Mat Sci 27: 2705
174. Krstic VD (1992) J Am Ceram Soc 75: 170
175. Cerovic L, Milonjic SK, Zee SP (1995) Ceramics Internat 21: 271
176. Martin HP, Muller E, Knoll Y, Strienitz R, Schuster G (1995) J Mater Sci Lett 14: 620
177. Martin HP, Ecke R, Muller E (1998) J Eur Ceram Soc 18: 1737
178. Lee JG, Butler IB (1975) Am Ceram Soc Bull 54: 195
179. Chawla KK (1998) Composite materials. Springer, Berlin, Heidelberg, New York, p 60
180. Krishnarao RV, Godkhindi MM, Ivengar PG, Chakraborty M (1991) J Am Ceram Soc
74: 2869
181. Krishnarao RV, Mahajan YR (1996) Ceram Internat 22: 353
182. Patel M, Padhi BK (1991) Indian J Technol 29: 9
183. Patel M, Padhi BK (1993) J Mat Sci Lett 12: 1234
184. Krishnarao RV, Mahajan YR (1996) J Mat Sci Lett 15: 232
185. Ul'yanov TM, Sivakova LT (1998) Russian J Appl Chem 71: 1450
186. Bohne Y, Martin HP, Muller E (1997) In: Scott ML (ed) Proceed. 11th ICCM, Vol 4.
Woodhead Publishing, p 643
187. Ota T, Takahashi M, Hibi T, Ozawa M, Suzuki H (1995) J Am Ceram Soc 78: 3409
188. Greil P, Lifka T, Kaindl A (1998) J Eur Ceram Soc 18: 1961-1975
189. Sieber H, Hoffmann C, Kaindl A, Greil P (2000) Adv Eng Mater 2: 105
190. Schulte K, Kainer KU (eds) (1999) Verbundwerkstoffe und Werkstoffverbunde, Wiley-
VCH Weinheim: a) Krenkel W, Hall N, Seiz S, p 301; b) Sieber H, Greil P, p 307;
c) Hoffmann C, Kaindl A, Sieber H, Greil P, p 393; d) Friedrich H, Sieber H, Greil P,
p 399
191. Sieber H, Kaindl A, Schwarze D, Werner JP, Greil P (2000) cfi/Ber DKG 77: 21
192. Greil P (1999) In: Kriegesmann J (ed) Technische Keramische Werkstoffe, Verl Dtsch
Wirtschaftsdienst Koln, Sect 4.3.2.1
193. Casady JB, Johnson RW (1996) Solid State Electron 39: 1409
194. Tairov YM, Tsvetkov VF (1981) J Cryst Growth 52: 146
195. Muller SG, Glass RC, Hobgood HM, Tsvetkov VF, Brady M, Henshall D, Jenny JR,
Malta D, Carter CH (2000) J Cryst Growth 211: 325
196. Augustine G, Balakrishna V, Brandt CD (2000) J Cryst Growth 211: 339
197. Ruff M, Hitlehner H, Helbig R (1994) IEEE Trans Electron Dev 41: 1040
198. Fissel A, Kaiser U, Ducke E, Schroter B, Richter W (1995) J Cryst Growth 154: 72
199. Fissel A (2000) J Cryst Growth 212: 438
200. Fissel A, Kaiser U, Pfennighaus K, Schroter B, Richter W (1996) Appl Phys Lett 68:1204
201. Fissel A, Pfennighaus K, Kaiser U, Schroter B, Richter W (1999) J Electron Mater 28:206
202. van Elsbergen V (1998) Thesis, University of Duisburg (Germany)
203. Schardt J (1999) Thesis, University of Erlangen-Niirnberg (Germany)
204. Tabata A, Kamijo H, Suzuoki Y, Mizutani T (1998) J Non-Cryst Sol 456: 227/230
205. Nakaaki I, Saito N, Inui Y, Yoshioka S, Nakamura S (1993) Philos Mag B68: 55
206. Gandia JJ, Gutierrez MT, Carabe J (1993) Sol Energy Mater Solar Cells 29: 139
207. Kumbhaar AA, Dusan RO, Bauer S, Schroder B (1998) J Non-Cryst Solids 452: 227/230
208. Guanghua C, Fangqing Z, Shaoguang Y (1991) J Non-Cryst Solids 1263: 137/138
209. Ichihara T, Aizawa K (1998) J Non-Cryst Solids 1345: 227/230
210. Chiu HT, Huang SC (1993) J Mater Sci Lett 12: 537
211. Demichelis F, Pirri CF, Tresso E (1993) Philosoph Mag B67: 331
212. Takeshita T, Kurata Y, Hasegawa S (1992) J Appl Phys 71: 5395
213. Giorgis F, Giuliani F, Pirri CF, Tresso E, Conde JP, Chu V (1998) J Non-Cryst Solids
227/230: 465
214. Senemaud C, Gheorghiu-de la Rocque A, Bonnefont PA (1998) J Non-Cryst Solids
447: 227/230
215. Watanabe A, Matsuda M (1991) Chem Lett 1101
216. Mucal MR, Milestone NB, Vickridge IC, Swain MV (1994) J Mater Sci 29: 4487
217. Scarlete M, Butler IS, Harrod F (1995) Chem Mater 7: 1214
218. Colombo P, Paulson TE, Pantano CG (1997) J Am Ceram Soc 80: 2333
219. Kajzar F, Messier J, Rosilio C (1986) J Appl Phys 60: 3040
220. Hayase S (1994) Chem Tech 24: 19
221. Furukawa S (1998) Thin Solid Films 33: 222
222. Ishida K, Sasaki D, Horiuchi T, Matsuchige K (1998) Nucl Instrum Methods Phys Res
B135: 270
223. Kani R, Nakano Y, Hayase S (1995) Macromolecules 28: 1773
224. Bunsell AR, Berger MH, Kelly A (1999) In: Bunsell AR, Berger MH (eds) Fine Ceramic
Fibers. Marcel Dekker, New York, Basel, p 27
225. Klein S (1999) Chemische Gasphasensynthese und Charakterisierung von nanokristal-
linem Siliziumkarbid-Pulver, Fortschr-Ber.VDI, Reihe 5, Nr546, Diisseldorf (Germany)
226. Lee HJ, Eguchi K, Yoshida T (1990) J Am Ceram Soc 73: 3356
227. Rao N, Micheel B, Hansen D, Fandrey C, Bench M, Girshik S, Heberlein J, McMurry P
(1995) J Mater Res 10: 2073
228. Mach R, Klotz HD, Oleszak F, Olschewski C, Dorfel I, Surke KD (1995) In: Galassi C
(ed) Fourth Euro-Ceramics, Vol 1 Basic science: development in processing of
advanced ceramics, Faenza Editrice (Italy) p 89
229. Fantoni R, Borsella E, Piccirillo S, Ceccato R, Enzo S (1990) J Mater Res 5: 143
230. Borsella E, Botti S, Alexandrescu R, Morjan I, Dikonimos-Makris T, Giorgi R, Martelli S
(1993) Mater Sci Eng A168: 177
231. Forster J, von Hosslin M, Uhlenbusch J (1996) Appl Phys B62: 609
232. Popp U, Herbig R, Michel G, Muller E, Oestreich C (1998) J Eur Ceram Soc 18: 1153
233. Muller E, Oestreich C, Michel G, Popp U, Staupendahl G (1999) In: British Ceramic
Proceedings, No 60, Vol 2, IOM Communications Ltd, London, p 191
234. Cardarelli F (2000) Materials handbook: 350. Springer, Berlin, Heidelberg, New York
235. Kriegesmann J (1999) Technische Keramische Werkstoffe 3: 32, DWD
236. Evans AG (1990) J Am Ceram Soc 73: 187
237. Grathwohl G, Kuntz M, Pippel E, Woltersdorf J (1994) Phys Stat Sol (a) 146: 393
238. Wilshire B, Carreno F (1999) Mater Sci Eng A 272: 38
239. Borsa CE, Ferreira HS, Kiminami RHGA (1999) J Eur Ceram Soc 19: 615
240. Descamps P, Sullivan O, Pooteman M, Descamps JC, Leriche A, Cambier F (1999) J Eur
Ceram Soc 19: 2475
241. Drissi-Habti M, Nakano K (1997) Comp Sci Technol 57: 1483
242. Park K, Vasilos T (1999) I Mater Sci 34: 2837
243. Cheong DS, Hwang KT, Kim CS (1999) J Am Ceram Soc 82: 981
244. Partridge G (1993) Adv Mater 5: 468
245. West G, Taplin DMR, Boccaccini AR, Plucknett K, Lewis MH (1996) Glastechn Ber
Glass Sci Technol 69: 34
246. Metcalfe BL, Donald IW, Bradley DJ (1993) Br Ceram Trans 92: 13
247. Mukherjee S, Ananth CR, Chandra N (1998) Composites 29A: 1213
248. Choy KL, Durodola JF, Derby B, Ruiz C (1995) Composites 26: 531
249. Martin HP, Muller E, Richter R, Roewer R (1997) Proc: 119, DGM Informationsge-
sellschaft, Frankfurt
250. Trabandt U, Knabe H, Rebstock K, Strobel F, Wulz HG (1997) Space Technol 17: 281
251. Schafer W, Vogel WD, Haug T, Friefi M (1995/1996) High Temp-High Press 321: 27/28
252. Greil P, Seibold M (1992) J Mater Sci 27: 1053
253. Mixhlratzer A (1999) cfi/Ber DKG 76: 30
254. Muhlratzer A (1999) Verbundwerkstoffe und Werkstoffverbunde: 470, WILEY-VCH,
Weinheim
255. Krenkel W, Hall N, Seiz S (1999) Verbundwerkstoffe und Werkstoffverbunde 301,
WILEY-VCH, Weinheim
256. Sieber H, Greil P (1999) Verbundwerkstoffe und Werkstoffverbunde: 307, WILEY-
VCH, Weinheim
257. Chawla KK (1998) In: Composite materials 164, Springer, Berlin, Heidelberg, New York
258. Leckie FA (1996) High-temperature structural materials (Cahn RW ed) 193, Chapman
and Hall, London
259. Yajima S (1975) Chem Lett 9: 931
260. Urano A (1999) Ceram Eng Sci Proc 20: 85
261. Serizawa H (1999) Ceram Eng Sci Proc 20: 443
262. Lipowitz J, Rabe JA (1997) Ceram Eng Sci Proc 18: 147
263. Sacks MD, Schliffele W (1998) Ceram Eng Sci Proc 19: 73
264. Hwang LR, Fergus JW (1996) Comp Sci Technol 56: 1341
265. Dietrich D, Stockel S, Weise K, Nestler K, Marx G (1999) Fresenius J Anal Chem 365: 255
266. Tressler RE (1999) Composites 30A: 429
267. More KL, Ailey KS, Lowden RA, Lin HT (1999) Composites 30A: 463
268. Fourvel P, Sylvestrini P, Rouillon MH, Vicens J, Gomina M (1990) J Mater Sci 25: 5163
269. Bestrand S, Droillard C, Pailler R, Bourrat X, Naslain R (2000) J Eur Ceram Soc 20: 1
270. Sun EJ, Nutt SR (1997) J Am Ceram Soc 80: 264
271. Dannemann KA, Mandell JF (1994) Mater Sci Eng A177: 95
272. Boccaccini AR, Strutt AJ, Vecchio KS et al (1998) Composites 29A: 1343
273. Murthy VSR, Jie L, Lewis MH (1989) Ceram Eng Sci Proc 10: 938
274. Vicens J, Doreau F, Chermant JL (1997) J Microsc 185: 168
275. Hahnel A, Pippel E, Schneider R, Woltersdorf J, Suttor D (1996) Composites 27A: 685
276. Pannhorst W, Beier W, Liebald R (1997) Verbundwerkstoffe und WerkstoffVerbun-
de:109, DGM-Informationsgesellschaft Verlag, Frankfurt
277. Rebstock K (1998) PhD Thesis, Erlangen
278. Gadow R, Speicher M (1999) Ceram Eng Sci Proc 20: 551
279. Gadow R, Kienzle A, Speicher M (2000) Fortschrittsberichte DKG 15: 135
280. Kaya H (1999) Comp Sci Technol 59: 861
281. Ohzawa Y, Nomura K, Sugiyama K (1998) Mater Sci Eng A255: 33
282. Singh JP, Sutaria M, Bielke W (1997) Ceram Eng Sci Proc 18: 719
283. Adler J, Standke G, Stoever H (1997) Handbuch Keramische Werkstoffe, Verlag DWD
284. Adler J, Heymer H, Standke G (1998) Werkstoffwoche 1998 (Proc): 239
Amorphous Multinary Ceramics in the Si-B-N-C System
Martin Jansen1, Britta Jaschke2, Thomas Jaschke3
Max-Planck-Institut fiir Festkorperforschung, HeisenbergstraBe 1, 70569 Stuttgart, Germany
1
e-mail: martin.jansen@fkf.mpg.de
2
e-mail: britta.jaeschke@fkf.mpg.de
3
e-mail: thomas.jaeschke@fkf.mpg.de
Inorganic random networks composed of silicon, boron, nitrogen, and carbon constitute a
novel class of ceramics with outstanding durability at elevated temperatures. The only
approach known to their synthesis is the pyrolysis of appropriate pre-ceramic polymers, which
on their part are either accessible via co-condensation of different molecular species, or by
polycondensation of single source precursors exhibiting structural features that are desired to
be present in the final ceramics. In this report the different synthetic routes known to date have
been compiled and critically assessed. Although the reliable knowledge of this rather young
class of materials is still limited, their fundamental features with respect to structures and
performance have already began to shine through. The range of their properties - mainly
depending on the synthetic route, but to some extent also on the composition - is quite
considerable. The upper limits for thermal loads with respect to the onset of weight loss
(decomposition) in an inert atmosphere, to the onset of crystallization and phase separation,
and to the resistance to oxidation in air, seem to be -2000 CC, -1900 °C, and -1500 °C,
respectively. Some representatives show unique combinations of properties, relevant for an
application at high temperatures, which clearly surpass those of SiC or Si3N4. For instance,
fibers made of SiBN3C, derived from the single source precursor Cl3Si(NH)BCl2, keeps its high
mechanical durability, as measured by creep resistance and tensile strength, up to
temperatures of -1400 °C. This is related to the fact that these amorphous fibers do not
show any grain boundaries and are virtually free of macroscopic pores. SiC, in contrast, is
crystalline and suffers from grain boundary sliding and subcritical crack propagation at
mechanical loading, at elevated temperatures. By now the feasibility of scaling up the synthesis
of SiBN3C to technical dimensions has been demonstrated by processing 150 kg-batches
(of polymer) which have been spun and pyrolyzed to ceramic long fibers.
Keywords: Si/B/N and Si/B/N/C systems, Single source precursors, Ceramic fibers, Ceramic
coating, Microstructuring
1 Introduction 138
2 Molecular Single Source Precursors 140

2.1 Introductory Remarks 140
2.2 Synthesis 141
2.3 Structural and Bonding Properties 145
3 Preceramic Polymers 152

3.2 Preceramic Polymers by Co-Condensation 154

138 M. Jansen • B. Jaschke • T. Jaschke
3.2.1 Polymers for Si/B/N Ceramics 154

3.2.2 Polymers for Si/B/C Ceramics 155
3.2.3 Polymers for Si/B/N/C Ceramics 156
3.3 Preceramic Polymers from Single Source Precursors 157
3.3.1 Linking Reagents 157
3.3.2 Polymers for Si/B/N, Si/B/C, and Si/B/N/C Ceramics 158
3.3.3 Structural and Physical Properties of Polymers 163
4 Pyrolysis 164
4.2 Experimental Procedures 165
4.3 Mechanistic Studies 166
5 Processing 168
5.2 Powder Processing, Coating, CVD, Infiltration,
Microstructuring 168
5.3 Fiber Spinning 171
6 Chemical and Physical Properties of Si/B/N/C Ceramics 172

6.2 Chemical Analyses 173
6.3 Structure Determination and Structural Features 174
6.3.1 Short Range Order 174
6.3.2 Medium and Long Range Order 175
6.3.3 Modeling of Si/B/N Random Networks 176
6.4 Bulk Properties of Polyborosilazane Derived Ceramics 177
7 SiBN3C-Fibers 181
7.1 Technical Performance 181
7.2 Comparative Assessment of SiBN3C- and SiC-Fibers 183
8 Concluding Remarks 185
9 References 187
1
Introduction
The class of oxide materials has greatly benefited from the chemical
exploration of multinary systems, virtually right from the beginning. Thus,
besides a number of binary compounds, mainly in use as structural ceramics,
there is an impressive plethora of multinary oxide materials being applied as
functional, as well as structural ceramics. In some instances these are even
Amorphous Multinary Ceramics in the Si-B-N-C System 139
top performers in their field, e.g. PZTs as ferroelectrics [1, 2] or the HTSC
cuprates [3-5]. These examples demonstrate at the same time how the
optimization of any desired properly by fine-tuning the chemical composition
greatly profits from the inclusion of multinary systems. For YBa2Cu307_x) the
maximum Tc is achieved when x is close to 0.4, while for x > 0.5 no transition
to the superconducting state is observed at all [5].
As a pronounced contrast, in the field of high performance non-oxide
ceramics, currently only binaries are in use. Thus, the manifold opportunities
for creating new and ever more capable nitride or carbide ceramics offered by
the use of multi-component systems seem to have remained essentially
unexplored. The main chemical reason for this lagging behind of non-oxide
ceramics are clearly the extremely low self-diffusion coefficients of silicon or
boron in their nitrides or carbides [6, 7]. Although the experimental data
available are rather limited, the numbers presented in Table 1 [8] suggest that
the temperatures needed to complete a solid state reaction between SiC and
Si3N4 in an acceptable length of time reach, or even exceed, the decomposition
temperature of at least one of the reactants.
This is in agreement with experiment, according to which no solid state
syntheses of multinary compounds within the system Si/B/N/C from the binary
border phases have been successful so far (for the corresponding thermody-
namic considerations cf. [9]). In the course of the past decades, however,
considerable progress has been achieved in the synthesis of multinary
compounds by applying an alternative synthetic approach via molecular and
polymeric intermediates (cf. Fig. 1).
This so-called 'polymer route' was introduced by Chantrell and Popper [10],
who proposed the use of inorganic polymers as starting materials for the
preparation of ceramics, opening up the rather ambitious perspective of easily
shaping monolithic 'green bodies' via this route. At the end of the 1960s,
Winter et al. [11, 12] pioneered such a process for the production of Si/C/N
fibers, and developed it to technical feasibility. Following the route developed
by Yajima et al. [13, 14], ceramic fibers (SiC, Nicalon) have been available on
Table 1. Self-diffusion coefficients of carbon, nitrogen, and silicon in Si3N4 and SiC [6-8]
Solute D o [1(T 4 m 2 s"1] T [K]
(self-diffusion coefficient)
Si (a-Si 3 N 4 ) 4.5 x 10~1 1673

1.0 x 1(T1 1773
N (a-Si 3 N 4 ) 1.2 x 10~1 1473-1683
Si (/?-SiC) 8.36 x 1(T 2283-2547
C (yS-SiC) 2.62 x 1(T 2128-2374
Molecular Polymerisation Prece ramie Pyrolysis Ceramic

precursors polymer material
Fig. 1. General synthetic approach to ceramics starting from molecular precursors

an industrial scale since the mid-1970s. In 1984, Seyferth and Wiseman

[15,16] improved and popularized the silazane route. These pioneering efforts,
and numerous papers following in their wake, have in common their aim to
produce crystalline, binary ceramics. However, the most recent and rather
successful approaches along the polymer route differ from the latter essentially
in two respects: they aim at (1) multinary and (2) amorphous ceramics [17-
20]. Thus far, the most promising results have been achieved in the quaternary
system Si/B/N/C. In this review we will try to cover the state of development
presently reached for that novel class of materials.
2
Molecular Single Source Precursors
2.1
Introductory Remarks
Molecular compounds of silicon and boron that can be employed as molecular

precursors for processing non-oxide ceramics along the polymer route have to
fulfill a number of conditions. In the first place they must bear functionalities
that allow them to be polymerized by any of the processes established in
chemistry, and they have to contain nitrogen and/or carbon, or at least they
should offer the opportunity for inserting nitrogen and carbon during
polymerization. Furthermore, as developments in the field of materials
research in general aim at large scale applications, the molecular precursors
should be easily available and inexpensive. General formulae for molecular
entities fulfilling the conditions mentioned are given in Fig. 2.
The ligands R can be exclusively halide, amide, hydride, or 'smart' leaving
groups like trimethylsilyl (TMS) or trimethylstannyl, or combinations of them.
Some, but never all, of the sites R may be occupied by saturated or unsaturated
organic substituents. In the case of boron hydrides, the Lewis base (D) adducts
are preferred because they are much easier to handle than the pure boranes.
The halide, amide, hydride, unsaturated hydrocarbon, TMS, and stannyl
functions can be employed for different kinds of polymerization reactions,
while alkyl and aryl block the site which they occupy for any such process.
Thus, by adjusting the number of hydrocarbon ligands, methyl being used
preferably, the degree of cross-linking and, consequently, the rheological
properties of the polymer can be tuned. Halides and alkyl-halides of silicon,
boron, phosphorus, and aluminum are easily available, in general. This holds
true for silicon in particular, since large scale industrial products are based on
silicon tetrachloride or on silicon alkyl or aryl chlorides.
R 2 -Si-R 4 x
R
R3 < 2
Fig. 2. General structures of molecular precursors for Si/B/N/C ceramics (for further
explanation see text)
Such molecular precursors, containing only one electropositive element, can

be used for the production of nitrides, carbides, as well as of carbide nitrides of
the respective elements. When these precursors are co-polymerized, quater-
nary or ternary ceramics, e.g. within the systems Si/B/N/C, Si/B/N, or Si/B/C
are also accessible. However, as the velocities of the substitution reactions
involved sometimes differ considerably for different molecules, spatial
inhomogeneities of the elemental distributions in the polymers almost
inevitably occur. Such inhomogeneities usually induce phase separation and
decompositions of the multinary ceramic into its binary border phases. It has
been shown that these effects can be circumvented by using single source
precursors. One and the same molecule contains the different metals or half
metals in the ratio desired for the final ceramic, linked via nitrogen and/or
carbon. Preferably, the backbone of such a single source precursor should
survive the full sequence of processing, and should be present as structural
units in the final solid material. For completeness, it should be mentioned that
the periphery of the single source precursor must enable polymerization too.
2.2
Synthesis
The multinary non-oxide ceramics considered here are not investigated for
primarily academic reasons but mainly for the purpose of future industrial
application. Thus, with regard to a sustainable and cost-effective production of
a given single source precursor on a large scale, the syntheses should not
involve costly steps. In particular, metallizations (Grignard reactions, lithia-
tions) must be avoided. In fact, meeting this point is challenging since in
current molecular chemistry metal organic intermediates play a key role in
running syntheses efficiently and well directed. These approaches are
affordable in the laboratory, but not yet on a technical scale. Nevertheless, it
can be reasonable to apply these advanced techniques of synthetic molecular
chemistry while screening new systems. However, as soon as one of the newly
found ceramics exhibits promising properties, the synthesis has to be adapted
to the demands of technical feasibility. In any case, for the development of
single source precursors intended to be utilized for mass production it is
indispensable to select appropriate starting materials which should be readily
available, easy to handle, and as cheap as possible.
The synthesis of a single source precursor can be laborious, and usually
requires several steps. Fortunately, for the formation of element-nitrogen
bonds a large stock of chemical tools developed in basic research can be
utilized [20]. In contrast, the options for establishing element-carbon bonds
are much more limited. Wurtz-analogous dehalogenations play a dominating
role, while for more special purposes hydroborations, salt eliminations, or
Miiller-Rochow type of reactions can be employed. However, as a fortuitous
circumstance, carbon from the substituents frequently inserts spontaneously
into the inorganic network upon pyrolysis.
The silazane cleavage offers a rather convenient and versatile access to
precursors containing a nitrogen bridge between the electropositive elements,
e.g., silicon and boron. This approach has been applied during the synthesis of
the, so far, most successful precursor (trichlorosilylamino)dichloroborane
(TADB). Starting from hexamethyldisilazane (HMDS), this single source
precursor is prepared in a two-step reaction with almost quantitative yields
[17, 21] (Eq. 1):
H H H
| +(H3C)nSiCI(4_n| | +BCI3 j
N ». ^,N^ m- N.
(H 3 C) 3 Si Si(CH 3 ) 3 -(CH3)3SiCI (H 3 C) 3 Si SiCI 3 _ n (CH 3 ) n -(CH3)3SiCI C I 2 B ^ ^ S i C I 3 n ( C H 3 ) n
n =0 TADB
n =1 MADB
n =2 DADB
n =3 TMDB
(1)
In order to reduce the degree of cross-linking during the successive

polymerizations, the chlorine atoms attached to silicon have been partly
substituted by methyl groups [22]. Along the same reaction sequence as shown
in Eq. (1), but using methyl or dimethyl chlorosilane as an educt, (meth-
yldichlorosilylamino)dichloroborane (MADB) and (dimethylchlorosilylami-
no)dichloroborane (DADB), respectively, were obtained in high yields. In the
first reaction step (Eq. 1) only one Si—N-bond in HMDS is cleaved by the
silane (H3C)nSiCl4_n. At mild conditions, the elimination of the second
trimethylsilyl group can only be achieved using the stronger Lewis acid BCl3.
As a further member of this family of precursors, (trimethylsilylamino)di-
chloroborane (TMDB) can be obtained by reaction of HMDS with BC13 at very
low temperatures [23].
A major advantage of this reaction path lies in the formation of a volatile
byproduct, i.e., chlorotrimethylsilane which can easily be distilled off.
Furthermore, the yields are usually high, and the heats of reaction low. Thus
reactions of this type can be scaled up without any 'in principle' problems. It
should be noticed that by reacting the byproduct chlorotrimethylsilane with
ammonia, the educt hexamethyldisilazane can be recovered, which is of
particular interest for ecological as well as for economic reasons.
For the formation of element nitrogen bonds, stannazane cleavage is a
widely applied alternative to the silazane cleavage. Since fission of the labile
Sn—N bond proceeds much more smoothly, stannazanes are preferred when
the reactivity of silazanes is not sufficiently high. A synthesis of a single source
precursor solely utilizing stannazane cleavages is shown in Eq. (2) [24, 25]:
Sn +SiCU (H 3 C) 3 Sn +2BC,3 CI2B
N—Sn(CH,), *- N—SiCI, *" N—SiCU
(H3C)3S/n -<CH3hSnCI (H^Sr, -2™C' Cl/
(2)
If the second step of this reaction is carried out with equimolar amounts
of BC13 and stannazane, the resulting dichloroboryl-trichlorosilyl-trimethyl-
stannylamine cannot be isolated as such, but dimerizes spontaneously [25] to a
cyclic precursor containing silicon and boron in a 1:1 ratio (Eq. 3):
CI CI
(H3C)3Sn
N-Sn,CH3)3 C I 2 B - l / ^N—B
- 2 ( CH 3 , 3 SnC,
CI3Si
CI CI
(3)
However, since organotin compounds are more or less toxic, depending on
their degree of alkylation, the production of a precursor on an industrial scale
using the stannazane route does not seem to be reasonable.
Recently, the well known dehydrohalogenation reaction has been employed
to open the access to another family of single source precursors [26]. Using a
gas phase reaction, dichloroboryl-methyl-trichlorosilyl-amine and dichloro-
boryl-dichloro(methyl)silyl-methyl-amine, respectively, can be produced in
high yields (Eq. 4). As this synthesis can be performed in a continuous process
and the appropriate educts are cheap, it is particularly well suited for a scale up:
C H CH
+ 4(H 3 C)NH 2 3 3
2 (H3C)nSiCI4_n r -
r
-2(H 3 C)NH 3 CI -(CH3)nCI3-nSiN(CH3)H2CI
n = 0 DMTA
n = 1 DDMA
(4)
Transmetallation is also appropriate for creating boron-nitrogen bonds.

Starting from molecules containing (acidic) NH functions, the nitrogen atom
is readily metallated by n-butyl lithium. Subsequent reaction of the lithium
amide with a suitable element halide provides the desired precursor, while the
lithium halide is eliminated (Eq. 5) [25, 27]:
CUSi
3 CloSi CUSi
3 \
\\ \
N
N
N—H N—Li
N N—BCI2 (5)
N
ci3s(( c3s( a3s(
For the formation of carbon-boron bonds basically two reaction types are of
importance: hydroboration and transmetallation (salt elimination). An elegant
route to alkyldihaloboranes has been introduced by Soundararajan and
Matteson [28,29], who employed a mixture of trichloroborane and trialkylsilane
to generate HBC12 in situ. This highly reactive intermediate hydroborates
alkenes readily and in high yields, even at low temperatures. In 1997 Jansen and
Jiingermann applied HBC12 for the hydroboration of vinylchlorosilanes, and
opened up a convenient access to a family of single source precursors containing
a (branched) carbon bridge between boron and silicon (Eq. 6) [30, 31]:
CH 3 TSDE
DSDE (6)
(CH 3 ) n CI 3 _ n Si (CH 3 ) n CI 3 _ n Si CSDE
Based on the results of Riccitiello et al. [32], who reported on the synthesis
of polyborosilanes through hydroboration of methylvinylsilanes by borane
adducts or diborane, Aldinger et al. [33], as well as Riedel et al. [34], applied
the hydroboration reaction of different borane adducts with various vinylsil-
anes obtaining single source precursors with a carbon bridge between silicon
and boron (Eq. 7):
Cl
C—Si—R (C2H4)—Si—R
| -D
Cl Cl (7)
R = H, CH 3 , Cl D = S(CH 3 ) 2 , N(CH 3 ) 3 , N(CH 2 CH 3 ) 3
Originally this approach had already been developed by Jones and Myers
about 30 years ago [35], who used the reaction of unsaturated organosilicon
compounds with diborane to produce silylorganoboranes. As a certain
drawback, mixtures of single source precursors are obtained along this route,
due to the lack of regiospecificity [35, 36]. Moreover, the regioisomers exist as
different diastereomers resulting in a complex mixture of molecules with
different chemical and physical properties. A workup procedure to isolate the
individual isomers prior to polymerization is supposed to be costly and has
not been tried, so far.
Another way to introduce carbon-boron bonds is the transmetallation
reaction, a rather versatile tool which can also be used to realize other element-
boron bonds [37].
The most widely employed procedure for the preparation of such
organoboranes is the well known Grignard reaction starting from a suitably
substituted halogenoalkylsilane and magnesium. Subsequent reaction of the
Grignard reagent with a haloborane leads to the desired organoborane (Eq. 8):
Cl Cl Cl ,
Mg C B( R)2 /
R—Si—CH2-CI > R—Si—CH2-Mg-CI ' ° » R—Si—CH 2 -B + MgCI2
X
ci ci ci °RI
(8)
^jl V/i in D ^^^ r\ ^j I— 1.1 |n ^j ' UCJLJI v i \ i
Unfortunately, the direct reaction of the silicon Grignard reagent with a

trihaloborane is hampered by severe obstacles. On the one hand the
preparation of the magnesium organyl can only be carried out using an ether
as solvent (e.g., diethyl ether or THF). On the other hand the trihaloborane
acts as a strong Lewis acid reacting rapidly with ethers, even at low
temperatures [38]. This conflict can be resolved by using a haloborane with
low Lewis acidity, e.g., a dialkoxyhaloborane. Subsequent halogenation of the
organoborane dialkylester yields the desired family of single source precursors
which are the first to contain a methylene bridge between boron and silicon
[39]. By reacting these precursors with hexamethyldisilazane the correspond-
ing borazine derivatives form in an almost quantitative yield (Eq. 9) [40]:
H
RCI 2 Si-CH 2 ^ ^ N v ^CH 2 SiCI 2 R
| /Cl + 3 (H3C)3Si-NH-Si(CH3)3 I I
3 R—Si-CH2-B^ • HN ^ N H
I Cl - 3 (H3C)3SiCI ? R = Cl TSMB
Cl
CH2 R = CH 3 DSMB
SiCI 2 R
(9)
A survey of all molecular single source precursors documented in the
literature up to the end of the year 2001 is given in Table 2.
2.3
Structural and Bonding Properties
The identities of all molecular precursors presented in Table 2 have been

confirmed by standard analytical techniques such as mass spectrometry and
NMR. Some of the precursors having melting points above ambient
temperature have been successfully crystallized, and their crystal structures
have been determined by X-ray diffraction. This has been achieved for some of
the species from Table 2: dimethylchlorosilyl-amino-dichloroborane (DADB)
(Eq. 1) [22], (trimethylsilylamino)dichloroborane [23], the two borazine
precursors TSMB and DSMB (Eq. 9) [41], the cyclodisilazane in Eq. (3) [25],
and the dimethylsulfide adduct of DSDE [42]. The molecular structures are
given in Fig. 3.
Unfortunately, those single source precursors which are farthest advanced
towards application are liquid at room temperature, and attempts to grow
crystals at low temperatures have failed, so far. However, reliable insights into
the relevant geometric/structural features have been gained by quantum
mechanic calculations. Some theoretical work concerning single source
precursors has been done by Marian et al., who combined experimental and
ab initio techniques to study the structural properties of l-(dichloroboryl)-
l-(trichlorosilyl)ethane (TSDE, cf. Eq. 6) and (trichlorosilylamino)dichloro-
borane (TADB, cf. Eq. 1) [43, 44].
The calculations show that for these molecules the respective conformations
as shown in Fig. 4 represent distinct and low lying minima at the potential
energy surface.
As expected, both structures reveal planar NBC12 and tetrahedral NSiCl3
units. In TADB, the planar Si-NH-BCl2 unit exhibits a partial N—Si and N—B
double bond corresponding to an N—B bond length of 142 pm, whereas in
TSDE all bonds have to be regarded as ordinary single bonds. Thus, the Si-N-B
backbone in TADB is reinforced stabilizing the molecule against fission during
polymerization and, to some extent, during pyrolysis which might be one
reason for the excellent overall performance of this precursor. All calculated
bond lengths computed with the B3LYP [45-47] density functional are in good
agreement with appropriate interatomic distances observed for the molecular
structures determined experimentally (Fig. 3a-f).
Besides calculations concerning the potential energy surface and bonding
parameters of TADB and TSDE, infrared spectra were computed and
Table 2. Survey of molecular single source precursors, derived polymers, and ceramics
Single source precursor Crosslinking reagent Composition of the Onset of weight References
ceramic loss/ c C
H Methylamine SiBN2.3C0.8 ("SiBN3C") 1755 [17, 21, 136]

I Ammonia Si3B3N7 1690
CI3Si "BCI,
(Trichlorosilylamino)dichloroborane (TADB)
H Methylamine SiBN2.8C,.7 1900 [22]

Ammonia SiB2.3N3.4 1575
(H3C)CI2Si BCI2
(Methyldichlorosilylamino)-dichloroborane (MADB)
H Methylamine SiB3.8N4.6C3.3 >2000 [22]
Ammonia 1437
(H3C)2CISi BCI2
(Dimethylchlorosilylamino)-dichloroborane (DADB)
H Not investigated Not determined Not investigated [23]
I
(H3C)3Si BCI2 3
to
(Trimethylsilylamino)-dichloroborane
H Cyanamide SiBN2C2 5 1671 [87]
Guanidine SiBN2.5C2 1465
Cyanguanidine SiBN2C0.4 1625
((H3C)2N)3Si ^B(N(CH 3 ) 2 ) 2
Formamidine 1358
Melamine Si3BN4C2 1812
Tris(dimethylamino)silylamino-bis(dimethylammo)-
borane (TABB)
H H Ammonia Not determined Not investigated [136]
N N
CI 3 Si X ^BCI X S i C I 3
Bis(trichlorosilylamino)-chloroborane (TACB)
CH 3 Methylamine Si2B2N5C4 1800 [30, 31, 85] •3
I
•JOI Dulo
l-(dichloroboryl)-l-(trichlorosilyl)-ethane (TSDE)
CH3 Not investigated Not determined Not investigated [30, 85]
(H 3 C)CI 2 Si / "BCIj
l-(dichloroboryl)-l-(dichloromethylsilyl)ethane (DSDE)
CH, Not investigated Not determined Not investigated [30, 85]
I
(H3C)2CISi "BCI2
l-(dichloroboryl)-l-(dichloromethylsilyl)ethane (CSDE)
CI,Si Methylamine Si2BC2N3 1850 [25]
3
\N—BCI
2
CI3Si
Dichloroboryl-bis(trichlorosilyl)amine
Table 2. (Contd)
ceramic loss/°C
Not investigated Not determined Not investigated [25]

((H3C)2N)2B Si B(N(CH3)2)2
N
I I
((H3C)2N)3Si Si(N(CH3)2)3
2)4-diboryl-B,B,B',B') 1,1,1,5,5,5-decakis-
(dimethylamino)-trisilazane
Methylamine >1900 [25]
X Ammonia Si3B6N10 Not investigated
N—SiCI3
CI2B
Bis(dichloroboryl)-trichlorosilyl-amine
Cl Cl Not investigated Not determined Not investigated [25]
\ /
Si
CI2B—N NN—BCI2
v/ \
Cl Cl
2,2,4,4-tetrachloro-l,3-bis(dichloroboryl)-
cydodisilazane
CH, 1870 [25]
CI3Si BCI2
Dichloroboryl-methyl-trichlorosilyl-amine (DMTA)
CH, Not investigated Not determined Not investigated [25]
(H3C)CI2Si ^BCI2
Dichloroboryl-dichloro(methyl)-silyl-niethyl-amine
(DDMA)
CH 2 Methylamine Si3B3N7C5 >1900 [39]
CI3Si "BCI 2
3
Trichlorosilyl-dichloroboryl-methane (TSDM)
5"
CH2 Methylamine Si3B3N5C7 >1900 [39]
(H3C)CI2Sii ^BCI
Dichloro(methyl)silyl-dichloroboryl-methane (DSDM)
H Methylamine Si3B3N7C4 >1900 [40]
CI3Si-CH2^ ^ N ^ ^Ch^-SiC^
B B
I I
HN. .NH
^B^
I
CH 2
SiCI3
B,B',B"-trichlorosilylmethyl-borazine (TSMB)
H Methylamine Si3B3N5C5 >2000 [40]
CI2(H3C)Si-CH2^ ^ N x / CH 2 -SiCI 2 (CH 3 )
B B
I I
HN. .NH
CH 2
SiCI2(CH3)
B,B',B"-dichloro(methyl)silyl-methyl-borazine (DSMB)
o
Table 2. (Contd)
ceramic loss/°C
Ammonia R = CH 3 : Si3.0BL0N2.0C43 >1900 [90]

R1 R = H: SijiBi.jN^Cic, >1900 [91]
-CH, Bis (trimethylsilyl) - R = Cl: not determined [33, 75]
L/li carbodiimide AY — V j n 3 . oix>o 41-^'2 42^*3 39 1600 [33, 75]
I R = H: not determined [33, 75]
Cl-Si-CI
I
R
R = Cl, CH3, H
R' = C2H4Si(R)CI2
Ammonia R = CH3: not determined Not investigated [91]
R' R = H: not determined Not investigated [91]
I
3
R"^ ^ChT
I
H—Si-H
I I
R
R = CH3, H
R1 = C2H4Si(R)H2
Si(Si(CH3)3)3 Hexamethyl-disilazane Si0^BNj 26C0_ Not investigated [88] >
1
HN NH
I I
B B
I
CH-, Ammonia Not determined 1350 [34, 42]

Cl
CH /
(H3C)CI2Si
Cl
l-(dichloroboryl)-l-(dichloromethylsilyl)ethane-
dimethylsulfide
I
Fig. 3a-f. Plots of molecular structures: a dimethylchlorosilyl-amino-dichloroborane;

b (trimethylsilylamino)dichloroborane; c tetrachloro-l,3-bis(dichloroboryl)-cyclodisilazane;
d l,3,5-tris(trichlorosilylmethyl)borazine; e l,3)5-tris(dichloromethylsilyl-methyl)borazine;
f l-(dichloroboryl)-l-(dichloromethylsilyl)ethane-dimethylsulfide
compared to the ones experimentally recorded. The agreement between

calculated and observed vibrations is very good [43, 44], which lends further
confidence to the structural data obtained from the calculations.
In the solid state, aminoboranes like (trimethylsilylamino)dichloroborane
and (dimefhylchlorosilylamino)dichloroborane aggregate to dimers forming
an almost square B2N2 four-membered ring (Fig. 3a, b). This tendency towards
dimerization has also been observed in solution: the nB-NMR spectra of both
compounds show two different boron signals which can unambiguously be
assigned to a three-coordinated and a four-coordinated boron species,
respectively. The relative intensities of the monomer's and the dimer's signals
vary with temperature, which is to be expected since the related dimerization
equilibrium is temperature dependent [22, 48].
Preceramic Polymers
3.1
As already mentioned, synthesis of multinary silicon boron nitrides or

carbon nitrides cannot be achieved via the well-known powder route,
including mixing, milling, and sintering of binary nitride/carbide powders,
because the interdiffusion of the covalent nitrides, and carbides, proceeds
H H CH 3
|i
Cl \l .Cl Cl \
2
£ % 3
3
( :i
2 \ c
Cl Cl
\
(
S
4
CI
Cl
3 5
1
TADB TSDE
Cl(2)Si-NH-BCl(4)Cl(5) lying in one plane Cl(2)Si-C(l)-B and C(l)-B-Cl(4)Cl(5) lying in one plane
Angles [deg]: Z Si-N-B 131.6 Angles [deg]: Z Si-C(l)-B 112.61

Z N-B-C1(4) 119.2 Z C(1)-B-C1(4) 121.48
Z N-B-C1(5) 121.4 Z C(1)-B-C1(5) 120.76
Z C1(4)-B-C1(5) 119.5 Z H-C(l)-C(2) 109.45
Z Cl(2)-Si-N 112.0 ZCl(2)-Si-C(l) 108.90
Distances [A]: N-Si 1.74 Distances [A]: C(l)-Si 1.893
N-B 1.42 C(l)-B 1.567
B-C1(4,5) 1.77 B-C1(4) 1.763
Si-Cl(l) 2.06 Si-Cl(l) 2.068
Fig. 4. Optimized geometry (B3LYP) for the single source precursors TADB and TSDE
extremely slowly. Thus, multinary nitride ceramics processed using powder

technology, up to now, are composites of the binary educts with a
microstructure largely determined by sizes and shapes of the grains in the
starting powders. Further drawbacks of powder processing are contamina-
tions caused by material losses of the milling tools or random milling
inhomogeneities. These problems can be overcome by using polymeric
precursors already containing the desired elements in a homogenous
distribution on an atomic level.
In principle, any polymerization technique established in chemistry can be
utilized in order to interconnect precursor molecules to polymers. In the case
of precursor molecules bearing olefinic groups, all reaction types developed
for the polymerization of olefins can be applied. However, since the leaving
groups at the inorganic molecules employed for producing ceramics should
preferably be halides, trimethylsilyl groups, or amides, the processes
predominating by far are aminolyses, including the special variants ammon-
olysis and transamination, and polycondensations as fast successive steps.
Less widely used for cross-linking are hydroboration and hydrogen elimina-
tion. Besides ammonia, a large variety of amines including multidentate
species, like guanidine, can by employed for the initial step, i.e., the
aminolysis. Overall, the polymerization step discussed here offers a consid-
erable number of sensitive processing parameters that can be varied over
rather wide ranges. While providing valuable options for tuning the process,
this plethora frequently makes it difficult to run the polymerization
reproducibly.
Quite generally, the polymerization stage in the polymer route to ceramic
materials plays a crucial role. Its effect on costs and ceramic yields is obvious,
as is its significance for the rheological properties of the polymer that influence
further processing. More importantly, the linking reagents selected, and the
choice of pressure and temperature during polymerization, widely determine
the structural features (crystalline, amorphous, homogenous) and thus the
overall performance of the final product.
With respect to the monomers used, two fundamentally different
approaches for generating preceramic polymers with several cationic species
can be distinguished. Either mixtures of molecules, each containing just
one electropositive element, are subjected to co-condensation, or single source
precursors which contain all the cations in the ratio needed for the final
ceramic are polymerized.
3.2
Preceramic Polymers by Co-Condensation
3.2.1
Polymers for Si/B/N Ceramics
In principle, the simplest way to produce preceramic polymers for ternary

silicon boron nitrides is to coammonolize mixtures of silicon and boron
chloride. Dietz has applied for a patent for such a process, with Si/B ratios
ranging from 9:1 to 1:9. There are two major disadvantages of this approach:
(1) the polymer is only accessible as a mixture with the by-product,
ammonium chloride, and (2) the ceramics obtained are composites constituted
of the binaries BN and Si3N4 [49].
Crosslinking of molecular units through hydrogen elimination represents
an attractive path, mainly because in this way the mass losses can be kept low.
As a crucial precondition, the precursor molecule and the crosslinking agent
have to bear differently polarized hydrogen atoms, i.e., protonic on one species
and hydridic on the other. An illustrative example is the reaction of
tetrakis(methylamino)silane with a Lewis-base adduct of borane, such as
BH3 • N(CH3)3, in an aprotic organic solvent. As an advantageous feature, this
synthetic route starts from easily available molecular educts, and the
compositions of the polymers have been tuned over a considerably wide
range of silicon/boron ratios from 10:1 to 1:1. Furthermore, properties such as
solubility, viscosity, or meltability are adjustable to the intended application
by treating the polymer thermally or by further crosslinking with ammonia or
methylamine [50].
Another route to polyborosilazanes, suitable for the synthesis of ceramics in
the Si/B/N system, consists of an adaptation of the well-known oxo-sol-gel
process to the ammonia system. Metal alkoxides are replaced by metal
alkylamides, and ammonia instead of water serves as the solvolysing agent.
Thus, mixtures of different metal alkylamides with comparable (!) velocities of

ammonolysis are reacted with ammonia to form polymetallasilazanes. Using
this approach, for instance, polyborosilazanes of any Si/B-ratio can be
obtained via co-ammonolysis of definite amounts of tetrakis(methylamino)
silane and tris(dimethylamino)borane [51-53].
A somewhat unconventional access to preceramic polymers containing
silicon and boron has been reported by Schmidt et al. A borazine polymer and
a silazane polymer are separately synthesized and dissolved in toluene, and
then mixed together. The subsequent crosslinking of the borazine with the
silazane components proceeds through hydrogen elimination and transamin-
ation reactions [54].
3.2.2
Polymers for Si/B/C Ceramics
Riccitiello et al. have synthesized preceramic polymers with Si—B bonds in their
backbones by a Wurtz analogous reaction of dialkyldichlorosilanes and boron
halides, either with or without adding methyl iodide for the control of molar
masses of the condensation products. Most of the polymers obtained are solid,
and soluble in hydrocarbons. Even though the Wurtz reaction is not specific, the
IR spectra of the polymer clearly indicate that Si—B bonds have formed
preferentially, but do not provide any evidence for the presence of Si—Si or B—B
bonds. Based on these results the authors suggest that the backbones generated
mainly consist of an alternating sequence of Si and B [55-59].
Furthermore, Riccitiello et al. have applied the hydroboration reaction to
preparing silicon boron carbon polymers. These polymers are generated by
reaction of a monomeric silicon compound exhibiting at least one carbon
double or triple bond, with diborane as well as various amine adducts of
borane. As silicon components, various vinyl- and ethinylsilanes can be
employed. When the borane ammonia adduct was used as a hydroboration
agent the polymers obtained contained small amounts of nitrogen. Regarding
the incorporation of nitrogen, it is assumed that part of the borane and
ammonia react with each other to form borazine, which then hydroborates the
unsaturated carbon-carbon bonds [60].
Riedel et al. have applied the hydroboration reaction for modifying
polymethylvinylsilanes with several borane adducts. The polymethylvinyl-
silane polymers were synthesized by a dechlorination reaction (Wurtz type
reaction) of dichlorodimethylsilane and dichloromethylvinylsilane with sodi-
um. The subsequent hydroboration of the vinyl groups has resulted in an
enhanced crosslinking of the polymers. The viscosity of the polyorganosilanes
can be tuned by varying the amount of borane as well as of vinyl groups in the
starting polymer [61].
Hemida et al. have prepared Si/B/C polymers by mixing polydimethylsilane
and the borane triethylamine adduct, followed by thermal treatment of the
mixture. Polydimethylsilane is assumed to undergo partially Kumada [62, 63]
rearrangements forming a polysilacarbosilane with Si-CH2-Si sequences and
reactive Si-H groups. The borane adduct acts as a crosslinking agent yielding
the final Si/B/C polymer. In order to increase the content of boron and the
extent of crosslinking, the polydimethylsilane is partly converted under
elevated pressure into a polysilacarbosilane, and subsequently again treated
with the borane [64].
3.2.3
Polymers for Si/B/N/C Ceramics
In general, modification or after treatment of a given silicon containing

polymer with an appropriate boron compound is a widely used method for the
preparation of polymers containing silicon and boron. Already in 1984,
Takamizawa et al. reported on the reaction of polydimethylsilane with
B-trimethyl-N-triphenylborazine. Thermal treatment of a mixture of these
educts results in decomposition and polycondensation reactions yielding an
organoborosilicon polymer [65, 66].
Seyferth and Plenio have applied the hydrogen elimination reaction to the
synthesis of polyborocarbosilazanes. As starting materials they have em-
ployed cyclic oligomers (CH3SiHN)n (n = 3-6) obtained from dichlorometh-
ylsilane and ammonia, and borane dimethylsulfide adduct. In accordance
with previous findings by Noth, who investigated the reaction of diborane
with hexamethyldisilazane, in a first reaction step cyclosilazane borane
adducts are formed and subsequently the B-N linkages are established
through hydrogen elimination, while the silazane rings of the oligomeric
silazane remain unchanged at first [67]. Eventually, borazine rings are
generated via ring-opening reactions at the cyclosilazane groups. Further
crosslinking takes place by the same process between the linear silazane
chains and the borane, and also through reactions among SiH- and NH-
functionalities. Polymers with various Si:B ratios are accessible by using
different (CH3SiHN)n:BH3 ratios. Si:B ratios as high as 20 and 10 result in
viscous oils, whereas larger amounts of BH3 (Si:B ratios of 2.5, 2, and 1)
produce solid polymers [68, 69].
A similar approach has been used by Zank employing three-dimensionally
crosslinked hydropolysilazanes. In spite of the steric hindrance, the reaction
between the resinous, highly branched polymers and the borane is reported to
proceed without destroying their structures and without further crosslinking,
virtually keeping their molecular weights unchanged [70].
Sneddon et al. have modified hydridopolysilazane polymers by dehydro-
coupling with borazine, B3N3H6, obtaining polyborocarbosilazanes in which
pendant borazine rings are bonded to the polysilazane backbone via B-N
linkages [71]. Due to reactive B-H groups, retained on the borazinyl rings,
further crosslinking can take place. Thus, this polymer does not seem suitable
for any processing that requires stable melt viscosities. To overcome this
disadvantage, Sneddon et al. have switched to boranes containing only one
B-H group, such as pinacolborane, l,3-dimethyl-l,3-diaza-2-bora-cyclopen-
tane, and 2,4-diethylborazine [72, 73].
Hydroboration reactions offer another access to boron modified silicon
polymers, as has been shown by Aldinger et al. who reacted different
polyvinylsilazanes [74, 75] and polyvinylsilylcarbodiimides [76-78] with

borane dimethylsulfide adduct.
Using miscellaneous boron sources, such as boron alkoxides, boron halides,
or aminoboranes, Funayama et al. modified various hydropolysilazanes.
Depending on the respective boron component used, different elimination
reactions take place, producing polysilazane chains with different boron-
containing groups attached [79-82].
As an alternative approach for generating homogeneous Si/B/N/C polymers,
co-condensation of smaller molecules appears to be more promising. Using
this route, Sneddon et al. have generated borazine/silazane backbone
copolymers by thermal co-condensation of the parent borazine with either
of two silazanes, tris(trimethylsilylamino)silane, and 1,1,3,3,5,5-hexamethylcy-
clotrisilazane. The polymerization occurs primarily through dehydrocoupling
providing B-N-Si-linkages between the borazine and the silazane. However, as
side reactions, silazane cleavages and homocondensation of the borazine also
take place. Therefore, the resulting polymer contains some repeating borazine
units or borazinyl rings attached to the copolymer backbone [73, 83, 84].
Another co-condensation reaction based on hydrogen elimination was
employed by Jiingermann and Jansen, using l,l,3,3-tetrakis(methylamino)-
2-methyl-2,l,3-azadisilolidine and borane trimethylamine adduct. As a char-
acteristic feature, the silicon-containing molecular educt comprises a carbon
bridge instead of the otherwise very common nitrogen bridge between two
silicon atoms. This offers an opportunity to increase the carbon content of the
resulting polymer, and of the ceramic [85].
A survey of the most relevant polymer syntheses by co-condensation is
given in Table 3.
3.3
Preceramic Polymers from Single Source Precursors
3.3.1
Linking Reagents
In principle, crosslinking of single source precursors can be achieved in the

same way as presented in Sect. 3.2 for the co-condensation of mixtures of
different educt molecules. However, compared to the co-condensation route,
the conditions for the single source precursor route are usually much better
defined, and a reaction control on a molecular level, at least during the initial
steps, seems to be possible. Thus, the linking reagents deserve more attention
when trying to tailor the polymers' and ceramics' properties. While the choice
in crosslinking agents is obviously restricted for hydroborations, a large
variety of nitrogen-containing compounds, all basically derivatives of ammo-
nia, is available. Choosing special amines instead of ammonia provides two
opportunities of major importance: (1) through the functionality of the
amine, primary or secondary, the degree of crosslinking, and thus the viscosity
of the resulting polymer, can be tuned; (2) the carbon content of the final
ceramic can be increased by using carbon rich amines.
All crosslinking species that have been employed up to now for the
synthesis of preceramic polymers are compiled in Table 2. Among these,
ammonia and methylamine are the most versatile, and thus these have been
preferably used. Ammonolyses or aminolyses, respectively, of chlorine
substituents at the single source precursors are the initial steps, which are
followed by condensation through deamination reactions. Using ammonia,
such polycondensation reactions proceed rapidly and provide highly cross-
linked polymers as solid, unsolvable, and unmeltable products, which with
certain respects is disadvantageous. Another drawback of using ammonia is
related to the formation of ammonium chloride as a byproduct, which is
difficult to separate from the polymer. The salt can be washed out using liquid
ammonia, a laborious process, or removed by sublimation, which is expensive.
Employing instead the bidentate methylamine results in polymers with a
significantly lower degree of crosslinking and lower molecular weights. The
oligomeric condensation products are liquid and can be easily separated from
the ammonium salt by filtration. Another advantage of aminolyses is that the
as obtained polymers can be tailored by an appropriate heat treatment to be
liquid, meltable, soluble, or unmeltable. Thus the polymers can be adjusted to
any desired application. At present, methylamine appears to be the most
attractive crosslinking reagent, and the carbon containing ceramics obtained
this way exhibit a very good overall performance [20].
Alkylamines bearing longer hydrocarbon chains show no improvement but
cause higher costs. In particular, including these chains in the preceramic
polymer is not a feasible route to increase the carbon content of the final
ceramic, since they facilitate undesired segregation of graphite [86, 87].
In some cases the less reactive hexamethyl disilazane represents an
attractive alternative to ammonia for crosslinking, because here the formation
of ammonium chloride as a byproduct can be avoided [88].
All crosslinking reagents mentioned so far generate nitrogen bridges
between the single source precursor molecules. In order to integrate carbon
into the main chain of the polymer and thus to vary the ratio of carbon to
nitrogen, alternative crosslinking reagents have been applied, which contain
carbon and at least two reactive amino- or imino groups, prerequisites that are
fulfilled by cyanamide, bis(trimethylsilyl)carbodiimide, formamidine, cyan-
guanidine, guanidine, or melamine. Since the latter agents are multifunctional,
crosslinking can be achieved not only by dehydrohalogenation but also by
hydrosilylation and hydroboration reactions (at the unsaturated carbon
nitrogen bonds) [86, 87].
3.3.2
Polymers for Si/B/N, Si/B/C, and Si/B/H/C Ceramics
Up to now the single source precursor route is mainly used for preparing
polymers in the quaternary Si/B/N/C system, and only to a minor degree in the
Si/B/N and Si/B/C systems. A synopsis of the reactions of the single source
precursors with respective crosslinking reagents is included in Table 2. The
most commonly used, and most versatile, procedure for synthesizing prece-
Table 3. Survey of polymers and derived ceramics obtained via co-condensation
Starting compounds Reaction type Composition of the ceramic Pyrolysis/ References

(molar ratios of the educts) °C
x SiCl4 + y BC13 + NH3 Ammonolysis 39.1% Si, 14.5% B, 45.2% N x:y = 1:1) 900 [49]
ulti
(x:y from 9:1 to 1:9) 23.1% Si, 26.2% B, 49.2% N (x:y = 1:3) 900
x Si(NHMe)4 + y BH3 • Nme 3 Dehydrogenation 1000 [50] i?
(x:y from 10:1 to 1:1) Si3B3N7(;c/=l:l) 1500 3
x Si(NHMe)4 + y B(NMe2)3 + NH 3 Ammonolysis 1000 [51-53]
IICS 1in
1500
t
Si^Ni! (*V= 2:1)

Si-polymer + B-polymer Dehydrogenation 51.7% Si, 2.1% B, 10.1% C (Si:B = 90:10) 900-1600 [54] if
T
Si-polymer: HSiCl3 + [(H3C)3Si]2NH 48.1% Si, 5.6% B, 8.1% C (Si:B = 75:25) 900-1600 00
B-polymer: (C1BNH)3 + [(H3C)3Si]2NH 32.9% Si, 16.7% B, 2.4% C (Si:B = 50:50) 900-1600 z
•
20.6% Si, 25.6% B, 0.8% C (Si:B = 25:75) 900-1600 1/1
x (H3C)2SiCl2 + y BC13 + z H3CI + Na Red. elimination 53.7% Si, 7.8% B, 30.3% C (x:y:z = 2:3:0) 1100 [55-57]
44.2% Si, 11.0% B, 32.0% C (x:y:z = 1:1:1) 1300 1
x (H3C)2SiCl2 + y (CH3)BBr2 + Na Not determined
x (C6H5)2SiCl2 + y (C6H5)BC12 + z Red. elimination Not determined [58, 59]
CH2Br2 + Na
x (C6H5)2SiCl2 + y BBr3 + z Not determined
CH2Br2 + Na
x (H3C)nSiR4.n + y H3B <- D (n = 0-3) Hydroboration Not determined 1000-1100 [60]
R = vinyl, ethinyl
D = NH3, N(C2H5)3, BH3
Polymethylvinylsilane + H3B <— D Hydroboration 48.8% Si, 6.7% B, 38.7% C 1000 [61]
D = S(CH3)2> THF 49.4% Si, 6.9% B, 39.4% C 1700
49.5% Si, 6.3% B, 39.9% C 2000
[(H3C)2Si]n + H3B <- N(C 2 H 5 ) 3 Kumada-rearrangement/ 35.1% Si, 3.6% B, 54.0% C 1000 [64]
dehydrogenation
o
Table 3. (Contd)
Starting compounds Reaction type Composition of the ceramic Pyrolysis/ References

(molar ratios of the educts) °C
[(H3C)2Si]n + [(H3C)BN(C6H5)]3 Thermal decomposition 60.5% Si, 0.9% B, 1.2% N, 34.7% C 1300 [65, 66]
[(H3C)Si(NH)]6 + H3B *- S(CH3)2 Dehydrogenation 45.3% Si, 5.4% B, 42.7% N, 0.3% C (Si:B = 4:1) 1500 [68, 69]
Hydropolysilazane + H3B <- THF Dehydrogenation Not determined [70]
Hydridopolysilazane (HPZ) + (HBNH)3 Dehydrogenation Si2.3B0.2N2.4CL0 (26.0 g HPZ + 21.68 g borazine) 1400 [71]
Si2.3B0.5N2.9CL0 (1.52 g HPZ + 3.24 g borazine) 1400
SiL9B2.6N4.gCL0 (1-51 g HPZ + 3.15 g borazine) 1400
Hydridopolysilazane (HPZ) + PIN-H Dehydrogenation Si1.00B0.10N0.95C0.51 (7.98 g HPZ + 8.00 g PIN-H) 1400 [72, 73]
(PIN-H: pinacolborane)
Hydridopolysilazane (HPZ) + Si1.00B0.04NLi0C0.50 (8.04 g HPZ + 8.29 g BCP-H) 1400
BCP-H(BCP-H: 1,3-dimethyl-l,
3-diaza-2-bora-cyclopentane)
Hydridopolysilazane (HPZ) + Si1.00B0.12N1.07C0.54 (7.96 g HPZ + 8.00 g DEP-H) 1400
DEP-H(DEP-H: 2,4-diethylborazine)
Hydroboration Not determined [74, 75]
T H| .TO 3 ' if H|
R
Jn LR J n
R = CH 3 , H, (NH) 0 . 5
H^C^CH2 ^B^ ,CH 3 Hydroboration Not determined [76-78]
+BH S C
-J4I_N=C=J- !- < H3>2 X|i_N=c=Nl
R = CH3, H, (NCN)o.5
[SiH2-NH]n + B(OCH3)3 Different elimination 50.5% Si, 6.8% B, 38.7% N, 0.6% C 1750 [79-82]
[SiH2-NH]n + B(NHC4H9)3 reactions (autoclave) 50.6% Si, 6.8% B, 39.3% N, 0.7% C 1750
[SiH2-N(CH3)]n + B(OC4H9)3 48.9% Si, 7.3% B, 38.2% N, 0.4% C 1750
x ((H3C)3SiNH)3SiH (TTS) + y (HBNH)3 Dehydrogenation Sioj9B1.00N1.o5C0.04 (xy = 20.7:38.3) 1400 [73, 83, 84] I
Sio.oiBi.ooN0.86C<0.oi (x-7 = 20.7:38.3) 1800
x ((H3C)2SiNH)3 (HCT) + y (HBNH)3 Sio.37B1.00N1.45C0.23 (x:y = 24.3:43.5) 1400
Sio.97Bi.0oN2.5sCo.30 (xy = 24.3:43.5) 1800 T"
Dehydrogenation Si 3 BN 5 C 5 1400 [85]

H3B—N(CH3)3
/ \
(H3C)HN N NH(CH3)
CH3
l,l,3,3-terakis(methylamino)-2-methyl-
2,1,3-azadisilolidone (TMMA)
ramie polymers starts with combining a solution of the respective single source
precursor and a cooled mixture of an excess of methylamine, or ammonia, and
hexane. This reaction mixture is then allowed to warm up to room temperature.
In the case of methylamine, the hydrochloride is filtered off and hexane is
removed from the polymer by vacuum distillation. Usually the polymer
obtained is a liquid of moderate viscosity. Through further heat treatment,
which must be carefully adapted to the respective polymer, the condensation of
the polymer can be continued until the desired viscosity is reached. TADB can
be processed using both ammonia and methylamine. Ammonolysis provides
an insoluble white solid, while the aminolysis yields AT-methylpolyborosilazane
(PBS-Me), which is liquid and soluble in hexane. It can be tailored by an
appropriate heat treatment to be liquid, meltable, soluble, or unmeltable. For
example, heating of the obtained PBS-Me in vacuum at 160 °C for about 2 h
gives a meltable and soluble polymeric borosilazane [17, 21].
Since crosslinking with cyanamide, formamidine, cyanguanidine,
guanidine, and melamine in any case leads to insoluble polymers, the
respective dimethylamide derivatives instead of fully chlorine functionalized
single source precursors have been used during explorative laboratory studies.
This requires the additional synthetic step of transforming TADB to
((CH3)2N)3Si-NH-B(N(CH3)2)2 (TABB). However, the latter precursor offers
a significant advantage since the byproduct dimethylamine is volatile and can
be easily separated from the polymers [86, 87]. So far, TADB and TABB are the
single source precursors to which the largest diversity of crosslinking reagents
has been applied.
Similarly, the single source precursors MADB and DADB can be polymer-
ized using ammonia and methylamine. MADB and DADB are different from
TADB in that one, or two, of the chlorine atoms, respectively, attached to
silicon are replaced by inert methyl groups. Due to the lower number of
chlorine functions available for polycondensation the resulting polymers show
a smaller degree of crosslinking than the TADB derived polymers. For example,
the polymer made from DADB and ammonia at room temperature is a soluble,
highly viscous liquid, whereas the corresponding polymer of TADB is an
insoluble solid. Accordingly, in addition to the already mentioned thermal after
treatment, there is another option for tuning the viscosities of the polymers by
varying the number of sites for crosslinking at the precursor molecules [22].
The precursors TSDM, DSDM, TSMB, DSMB, and [(Cl2B)2NSiCl3]2 have
been polymerized with both ammonia and methylamine, whereas DMTA,
TSDE, and Cl2BN(SiCl3)2 have only been reacted with the most promising
crosslinking reagent methylamine, up to now (Table 2) [25, 31, 41].
Polycondensation of tris[(dichloromethylsilyl)ethyl]borane with ammonia
results in an organopolyborosilazane with 4 wt% boron content. Again, the
nasty disadvantage of this procedure is the separation of the polymer from the
byproduct ammonium chloride [89, 90].
Employing bis(trimethylsilyl)carbodiimide instead of ammonia elegantly
leads to boron modified polysilylcarbodiimides. This has been found when
using the single source precursors, tris[(dichloromethylsilyl)ethyl]borane,
as well as its derivatives tris[(dichlorosilyl)ethyl]borane, and
tris[(trichlorosilyl)ethyl]borane. Again, the variation of the substitution

patterns of the single source precursors results in polymers exhibiting
significantly different amounts of crosslinking [33, 75].
Borazine-based Si/B/N/C polymers have been obtained by polycondensing
H3N3B3(Cl)2Si(SiMe3)3 with hexamethyldisilazane [88]. Here once again
silazane cleavage, forming the volatile chlorotrimethylsilane as a byproduct,
is the driving process.
An example for a conversion of a single source precursor to a preceramic
polymer through dehydrogenation is given by the base catalyzed (rc-butyl
lithium) dehydrogenative coupling of tris(hydridosilylethyl)boranes and
ammonia [91].
3.3.3
Structural and Physical Properties of Polymers
In general, all preceramic polymers are susceptible to further crosslinking due

to ongoing polycondensation, even at ambient conditions. Thus, this kind of
polymer can be denoted "living" polymers. This issue is crucial to all aspects
related to storage, handling, and processing as well as to analytical
characterizations.
So far, only few Si/B/N/C based polymers have been investigated in detail.
The most extensive data exist for the TADB derived polymer PBS-Me. Reaction
of TADB with methylamine and subsequent heating in vacuum at 160 °C results
in a soluble and meltable polymeric AT-methyrpolyborosilazane with about 80%
yield. Taking into account the presence of bridging and terminal amino groups
the empirical formula can be written as [(Si-NH-B)(NCH3)x(NHCH3)y]n,
2x + y = 5. Considering the composition as determined by chemical analysis
(Si 17.9%, B 6.9%, N 38.6%, C 27.5%, H 7.2%, O 0.7%, Cl 0.06%), x and y turn
out to be 1.7 and 1.6, respectively. Thus, the polymer includes virtually equal
amounts of bridging and terminal methylamino groups [17, 92]. Annealing
causes a continuous increase of bridging units. Figure 5 shows a schematic
drawing of the structural model, which is in agreement with the chemical
analyses, and IR as well as MAS NMR spectroscopy measurements. The FT-IR
spectrum shows the characteristic stretching and bending vibrations for all
functional groups expected to be constituent units of the polymer, i.e., v(NH)
H3C
N \ 'CH3 PH. %
3 N
CH3 " ° " \ X""*S / ~-H W"
Si .N~rn NH
CH3
,SU / \ ""'"—Sk^ /CH3
/ V ^ / I IM M \ NH
NH C N H
.NH -=v vr* ^ 1 "" —
NH
Fig. 5. Schematic drawing of the polymeric network of PBS-Me

(3414 cm" 1 ), v(CH3) (2933, 2886, 2808 cm" 1 ), <5(CH3) (1496, 1357 cm" 1 ), v(BN)
(1196 cm" 1 ), v(CN) (1080 cm" 1 ), v(SiN) (1050, 929 cm" 1 ), and <5(SiN) (768,
695 cm" 1 ). The 29Si MAS NMR spectrum exhibits one relatively sharp signal
with a half width of about 500 Hz at -33.5 ppm, which is the typical chemical
shift for silicon in a tetrahedral coordination by nitrogen. Finally, the n B MAS
NMR spectrum exhibits a signal at 13.3 ppm with a shape characteristic for
trigonal planar BN3-units [92].
The feature, which is crucial for the rheological properties of a polymer, is
its molecular mass distribution. This quantity has been determined for PBS-
Me by gel permeation chromatography in the range between 100 and
50,000 g mol" 1 (Fig. 6). Samples with viscosities from 4 to 290 Pa s have been
studied. While those samples with low viscosity (4-50 Pa s) contain predom-
inantly oligomers with relatively low molar masses of about 2000 g mol" 1 , the
more viscous samples (70-290 Pa s) show a bimodal distribution; in addition
to the peak corresponding to molar masses up to 2000 g mol" 1 , a second
peak for masses above 50,000 g mol" 1 appears. The latter peak grows with
increasing viscosity of the polymer, indicating that the fraction of the heavier
oligomers rises at the expense of the lighter oligomers [48].
For the polymers derived by polymerization with methylamine from TADB,
MADB, and DADB, respectively, rheological studies have been performed.
While the TADB-based polymer exhibits viscoelastic behavior, with the
viscous contribution being ~10 times higher than the elastic one, the MADB-
and DADB-based polymers show an almost purely viscous behavior. For the
latter two polymers, the ratio of the viscous and the elastic contributions
amounts to ~100. Furthermore, it has been found that the TADB-based
polymer shows a higher tendency to further crosslinking than the MADB- and
DADB-based polymers [93].
4
Pyrolysis
4.1
It appears reasonable to distinguish the steps of polymerization and pyrolysis

conceptually, even though the transition from the molecule to the ceramic via
oligomeric and macromolecular intermediates proceeds continuously. Usually,
after the polymerization, the polymer undergoes additional thermal treatment,
during which further polycondensation takes place. Upon raising the
temperature to several hundred degrees, fragmentation of the organic
constituents begins to mark the transition to 'pyrolysis'. During pyrolysis all
volatile components (mainly organics) are expelled, until a pure inorganic
solid residuum is obtained. These sequential processes proceed through
numerous, usually metastable and frequently ill denned, intermediates. Up to
now, one is still far away from understanding the atomistic mechanisms
involved in the pyrolysis of preceramic polymers. However, by applying
different ex situ and in situ probes during the pyrolysis, the gaseous
0.00 1.50 3.00 4.50 6.00 7.50 9.00 10.50 12.00

retention time / min
Fig. 6. Gel permeation chromatograms of PBS-Me samples with different viscosities

(column: 500 A divinylbenzene-styrene; eluenfctetrahydrofurane 1.5 ml min"1)
compounds evolved and the densification reactions have been monitored for
some selected polymers. This empirical knowledge has allowed the optimi-
zation of the process, and increased its reproducibility.
4.2
Experimental Procedures
Since all preceramic polymers for Si/B/N/C ceramics are extremely sensitive to
oxygen and moisture, they have to be handled under strictly inert conditions.
166 M. Jansen • B. Jiischke • T. Jaschke
This holds true, in particular, for the stage of pyrolysis, during which oxygen
contaminations of up to 10% may occur. Since oxygen impurities exceeding
1 wt% significantly affect the performance of the ceramic product this issue is
of crucial importance.
On a laboratory scale, the pyrolyses are performed in horizontal furnaces
equipped with quartz or gas-tight corundum tubes. The most relevant
experimental parameters are temperature, with special emphasis on the
temperature program applied, pressure, and the type of atmosphere. Two
different classes of gas atmospheres have been employed: inert gases,
commonly argon or nitrogen, and reactive gases. In the former case the
flowing gas merely acts as a carrier for removing the volatile decomposition
products, while in the latter it takes part in the reaction itself. For nitrides, the
reaction gases preferably used are ammonia and nitrogen. Ammonia is of
primary importance in the synthesis of pure nitrides from carbon-containing
precursors, since it removes carbon very efficiently. Apparently NH3 works as
an aminizing or transaminizing agent, substituting alkylamines, and enhances
crosslinking.
A typical temperature program includes three steps: (1) heating to the
desired pyrolysis temperature at a rate of 3-10 K min" 1 , (2) remaining for a
dwell time at the maximum temperature, and (3) cooling the sample to room
temperature with a rate of 5 K min" 1 [20].
4.3
Mechanistic Studies
Since all polymeric intermediates, and in many instances also the final
ceramics, are amorphous, only thermal and spectroscopic methods can be
utilized to characterize the thermal conversion. The most extensive studies
have been performed on the polymer W-methylpolyborosilazane (PBS-Me),
made from the single source precursor TADB. The pyrolysis has been
monitored in situ by differential thermal analysis combined with thermo-
gravimetric analysis and mass spectrometry (DTA/TG/MS). For ex situ
investigations, batches of the polymer were treated at different temperatures,
cooled to room temperature, and characterized by infrared spectroscopy and
nuclear magnetic resonance spectroscopy.
The thermal degradation, as monitored by DTA/TG/MS, passes through
three well resolved stages. From 200 to 500 °C the polycondensation,
accompanied by methylamine elimination, is completed (weight loss approx.
12%). In the second step, at about 600 °C, the biggest weight loss (approx.
28%) occurs. Mostly methane and hydrogen, but also hydrogen cyanide,
methylamine, acetonitrile, and dimethylamine are emitted. Another inflection
point of the TGA plot occurs at about 1100 °C. The weight loss is rather low
(approx. 4%), and is attributed to the loss of residual hydrogen, and small
amounts of nitrogen [20, 92].
Similar investigations on some other polymers have been reported.
Particularly, the second step of the pyrolysis at 600 °C, which is characterized
by the release of mainly methane and ammonia, also occurs for other polymers
such as polysilazanes [94-96] and polyborosilazanes [50, 68, 83, 97] produced
by co-condensation reactions. In contrast, the decomposition of hydridopo-
lysilazanes does not proceed in such well resolved distinct steps, but virtually
continuously within the temperature range from 200 to 800 °C [71].
Depending on the nature of the polymer, condensations can proceed via
hydrogen elimination [71], release of ammonia [96-100], or elimination of
trimethylsilyl groups [83]. In those few cases where the investigations were
extended to temperatures above 1000 °C, always some release of hydrogen was
observed at 1100 °C [96,97].
For the ex situ investigations of the pyrolysis, a series of FT-IR spectra were
recorded on samples of PBS-Me, which had been pyrolyzed at temperature
intervals of 100 °C between 200 and 1100 °C. Up to 600 °C no significant
changes were noticeable; the characteristic functional groups of the polymer
remaining virtually unchanged. At 600 °C the IR-spectra indicate fundamental
changes in the constitution of the samples. The intensity of C-H stretching
vibrations decreases dramatically and, instead of numerous sharp absorption
lines, broad bands appear in the region for nitrogen-element vibrations. Most
significantly, a vibration related to a Si-H stretching mode at about 2200 cm" 1
emerges. As the pyrolysis proceeds this peak disappears at about 1000 °C,
along with all other element-hydrogen vibrations [126, 128]. The 29Si MAS
NMR spectra recorded on the same samples all show one broad signal, which
corresponds to a chemical shift typical for silicon in a tetrahedral coordination
by nitrogen. At about 600 °C the resonance shifts from -34 ppm to -41 ppm,
which is due to changes in the second coordination sphere. Remarkably, the
half width of the signal grows from 500 up to 2000 Hz during the densification
and curing of the network. This can be related to two effects: a decrease in
mobility and an increase in strain. The latter causes deformations of the
characteristic surroundings of the constituting elements in the course of the
transformation of the polymer into a fully inorganic, amorphous material.
That kind of inhomogeneous line broadening has to be attributed to angular
distortions rather than to a spread in Si-N distances which according to the
pair distribution function (PDF) is surprisingly small [126-128]. The marked
increase of the half width of the signals in 29Si NMR spectra during pyrolysis
has been observed for all polymer derived Si/B/N/C materials investigated so
far, and appears to be a common feature of these systems [101, 102]. Over the
whole range of the thermal degradation, the n B MAS spectra exhibit one signal
characterized by a shift and lineshape typical for boron coordinated trigonally
planar by three nitrogen atoms. As compared to the n B MAS signals of the
fully pyrolyzed samples, which exactly match those of h-BN, the boron signal
of the polymer exhibits some deviation from this typical lineshape due to a
more irregular environment of the boron atoms.
The so-called 'ceramic yield' is a very problematic issue, from two points of
view. First, the economic performance of a ceramic processed along the
polymer route is strongly determined by the availability (i.e., price) of the
starting materials and by the degree to which they can be converted to the final
product. Second, during pyrolysis the organic fragments separate in a gaseous
state, and the size and number of macroscopic pores or cracks depend on the
amount of the mass loss, and in order to obtain ceramics as dense as possible
the mass loss has to be kept low. Unfortunately, the issue of ceramic yield is, as
a rule, not addressed properly in the literature. In particular, the reference
points for calculating the mass balance are not defined uniformly, mostly not
at all. It would be desirable to have two numbers characterizing the material
losses, one referring to the economic, and the other one to the processing
aspects, respectively. In the first the whole input of material should be
balanced against the yield in product and the reusable byproducts. This has
been done for Si3B3N7 and SiBN3C: 8.5 kg primary materials are required for
1 kg Si3B3N7, 6.8 kg byproduct can be recycled, and during pyrolysis 0.7 kg
are lost. The corresponding values for SiBN3C are: 7.0 kg starting materials
yield 1 kg ceramic, 5.3 kg recyclable waste, and 0.7 kg loss upon pyrolysis
[20]. Regarding the second aspect, it is much harder to define a proper
reference point for balancing the losses during pyrolysis. To us, as the only
viable possibility, it appears to be reasonable to define a certain value for the
viscosity, and use the masses of polymers at the stage of that very viscosity as
references.
5
Processing
5.1
Since the preceramic molecules and polymers under discussion are available in
virtually any physical state a wide variety of options is available for their
processing. Gaseous or volatile single source precursors can be employed for
chemical vapor deposition, liquid or meltable preceramic polymers can be
used for injection molding, coating, infiltration, and fiber spinning, and,
finally, solid precursors and the derived ceramics can be processed via the
conventional powder route. In particular in the liquid state the full range of
technologies developed in classical polymer chemistry can be applied to the
preceramic polymers. Although, up to now, only a few of those possibilities
have been explored in depth, the extremely promising potential of this novel
class of materials clearly has become apparent, also with respect to processes
leading to a broad variety of shapes.
5.2
Powder Processing, Coating, CVD, Infiltration, Microstructuring
In this section some as yet only sporadically investigated approaches to

generate defined shapes are summarized. As a consequence of their high
thermal durability, Si/B/N/C-based ceramics are extremely difficult to sinter.
Thus, compacting powders to workpieces without using sintering additives has
not yet been achieved, and no efforts aiming at the development of appropriate
sintering aids are documented up to now. As an alternative approach, highly
cross-linked polymer powders can be compacted and pyrolyzed [103]. In order
to decrease the porosity of ceramic workpieces, mixtures of two powders that

had been subjected to different temperature treatments have been used.
Pyrolysis at 1400 °C of a pre-compacted mixture of pristine PBS-Me and PBS-
Me treated at 500 °C yields ceramic workpieces of relative densities of up to
80%. Here the presence of the reactive polymer is necessary to obtain the
interconnection of the particles (essentially serving as a sintering aid) [104].
The difficulties encountered in the production of bulk pieces are becoming
irrelevant when reducing the dimensions. Protective coatings of SiBN3C on
carbon materials and on metals are of significant technical importance because
of their thermal stability and oxidation resistance. In these respects, SiBN3C
coatings are clearly superior to SiC used currently in order to allow the use of
carbon-containing materials above a temperature limit of about 500 °C. Since,
moreover, the coefficients of thermal expansion of SiBN3C and carbon are
comparable, and thus only low thermal tension between the coating and the
carbon substrate is to be expected, SiBN3C coatings are particularly suitable
for carbon containing materials. The potential of SiBN3C as protective layers
has been demonstrated by experimental coatings of, e.g., graphite [20], carbon
fiber reinforced composite materials (C/C and C/SiC) [105], carbon fibers
[106], and some metals [107, 108]. Here, different methods for the fabrication
of the ceramic coatings have been used. The coating of workpieces can be
performed by dip coating, either in a slurry of fine SiBN3C powder and molten
PBS-Me or in a 40 vol.% solution of the polymer in an organic solvent. Careful
pyrolysis at 1600-1800 °C in an oxygen free atmosphere yields a dense, crack-
free coating. Depending on the desired thickness of the coating, this procedure
can be repeated several times [103]. Thus, an SiBN3C-coated carbon cube
exposed to an oxyhydrogen flame could not be inflamed, showing an
impressive high thermal stability.
In addition to the extremely low cost dip coating process, the well
established, but more expensive, CVD (chemical vapor deposition) methods
can be applied for the preparation of SiBN3C coatings. The CVD process has
been used for coating carbon fibers using PBS-Me as precursor [106]. For this
purpose a PBS-Me fraction was prepared, which can be evaporated at
moderate temperatures and at low pressure. Depending on the deposition
time, coating thicknesses of between 1 and 5 um have been achieved. The
coating shows high mechanical strength, good adhesion, and provides by far
the best oxidation resistance of any coatings these authors had studied before.
For coating a CFC (carbon fiber reinforced carbon matrix) workpiece via
CVD, the monomer TABB has been used as the volatile precursor [105]. The
layers one obtains are dense and do not exhibit any grain boundaries.
Furthermore, large pores in the substrate surface (>1 um) are filled.
In order to improve the wear resistance of machining tools made of hard
metals, steel, tungsten carbide cobalt, titanium aluminide, titanium and silicon
substrates have been coated with SiBN3C via CVD of TABB [107, 108]. Layers
produced in this fashion are dense and exhibit a Vickers hardness of 1000-
2300 kg mm" 1 , depending on the specific process parameters.
Due to its solubility and the adjustable viscosity, PBS-Me is suitable for the
polymer infiltration process used in the production of CMCs (ceramic matrix
composite). Workpieces have been produced by infiltration of porous fiber

preforms using the RTM (resin transfer molding) process [105]. Compared
with other infiltration methods such as CVI (chemical vapor infiltration), the
polymer infiltration/pyrolysis technique offers distinct advantages: the
procedure is very simple and economic, and since the processing conditions
are rather mild, the fibers will not be damaged, neither chemically nor
mechanically. In particular for the production of carbon fiber reinforced
composites with a ceramic matrix, polymer infiltration represents an already
widely used method. So far, polycarbosilanes and polysilazanes are prefer-
entially employed for producing composite materials [109], although the
SiC- and SiCN-matrix composites suffer from both a rather high tendency
to crystallize and a merely moderate resistance against oxidation. These
disadvantages can be overcome by using PBS-Me as the preceramic polymer.
In general, the pores and cracks included in the infiltrated parts amount to a
rather high total porosity of 10-17%, regardless of the element combination
employed [109, 110]. Using PBS-Me, porosities of between 5 and 10 vol.% can
be reached after 4-5 infiltration cycles (for samples containing approximately
50 vol.% fibers) [105]. Recently another polymer for the Si/B/N/C system has
been used for the production of carbon fiber reinforced CMCs employing
similar fabrication techniques [111]. Here, carbon fiber woven fabrics
(Torayca) were infiltrated and pyrolyzed on a laboratory scale yielding
CMCs with porosities of 15% after three infiltration cycles. However, no
measurements of the mechanical properties of the workpieces have been
performed.
A preceramic polymer made from TSDE and methylamine has recently been
employed by Whitesides et al. for the generation of Si/B/N/C microstructures
and components for microelectromechanical systems (MEMS) using a special
set of techniques called 'soft lithography' [112]. Because of the high durability
of Si/B/N/C ceramics, which was confirmed in the course of the latter
investigations, microstructures made of Si/B/N/C can serve particularly well in
devices that have to withstand harsh oxidative and thermal environments,
such as microcombustors, microturbines, high-temperature sensors and
actuators. The fabrication of these micrometer sized surface features is
achieved by filling a polydimethylsiloxane mold with melted preceramic
polymer and subsequent curing at 200 °C for ~ 1 h. Upon increasing the
temperature slowly (=2 K min" 1 ) to >1050 °C, gas evolution occurs slowly
enough to prevent the ceramic from forming large voids and cracks, resulting
in a smooth and nearly perfect surface of the microstructure. During the
conversion of the polymeric structure into the final ceramic, lateral contrac-
tion of ~30% occurs due to the release of low molecular weight gases, e.g.,
methylamine, methane, and hydrocyane. Figure 7 shows an SEM image of a
hexagonal grid as well as a micrometer-sized gear of Si/B/N/C ceramic made
by using soft lithography.
In principle, all meltable preceramic (Si/B/N/C) polymers are suitable for
the fabrication of lightweight microstructures using the soft lithography
process. However, the evolution of gas during pyrolysis must not exceed a
limit value to avoid the formation of macroscopic defects.
250 \im
Fig. 7a, b. SEM images of: a hexagonal grid; b a free-standing micrometer-sized gear of
Si/B/N/C ceramic [112]
5.3
Fiber Spinning
Because of the high ceramic losses, a disadvantageous feature common to all
polymer routes, the most promising applications to start with are those aiming
at parts of small dimensions. Generally, films or fibers can be produced free of
macroscopic pores and cracks as long as the thicknesses do not exceed a few
tenths of a millimeter.
The combined properties of SiBN3C make it ideal for the production of
ceramic fibers. Since the preceramic polymer PBS-Me is extremely sensitive to
oxygen and moisture, all processing steps, fiber spinning, curing, and
pyrolysis, have to be carried out in an inert gas atmosphere. Fortunately,
the preceramic polymer PBS-Me can be processed via conventional polymer
techniques, either by melt-spinning or by a dry-spin process. Melt-spinning is
preferable because no solvent is required and, in general, the resulting "green"
fibers show better properties. Usually, a continuous melt-spinning process
comprised of spinning and drawing is used. After melting, and filtering, the
molten polymer is extruded through a spinneret with more than 300 holes at
temperatures of ~ 150 °C into a spinning funnel where the melt solidifies into a
filament form [113, 114]. The viscosity of the PBS-Me polymer can easily be
adjusted to the desired working range of 80-260 Pa s at shear rates between
400 and 1000 Pa s. Under these conditions, ceramic fibers with lengths
exceeding 200 m and diameters larger than 15 urn can be produced. Within
this range of shear rates, the polymer essentially behaves like a Newtonian
fluid. For production of fibers with diameters less than 15 um, smaller
capillary diameters and higher shear rates (up to 3200 Pa s) are required.
Curing of the (thermoplastic) green fibers can be achieved either thermally
or chemically. As thermal curing is slow for kinetic reasons (removal of
methylamine), chemical curing with gaseous species is preferred. Besides
curing of batches of the green fibers, in situ curing directly in the spinning
funnel is advantageous. By applying, for example, HSiCl3, curing of the green

fibers with diameters of ~20 urn is completed within a few seconds.
Since the velocities for spinning and curing of the green fiber (45-
100 m min"1) and pyrolysis (1 m min"1) are extremely different, a fully
continuous process is not yet feasible and the cured green fibers have to be
deposited before pyrolysis. The sufficiently cured green fibers are pyrolyzed by
exposing them to increasingly high temperatures, starting at ambient temper-
ature and ending up at about 1500 °C. During this process, all organic species
are removed and, in a yield of 55 wt%, purely inorganic SiBN3C fibers are
obtained. The ceramic yield is of importance not only from an economic point of
view. If released over a narrow temperature range, the gaseous byproducts can
cause damage (cracking) to, or even rupture of, the ceramic part. Fortunately,
during pyrolysis of cured PBS-Me fibers, the pyrolytic gases leave the system
over the rather wide temperature range (300-800 °C), and the integrity of the
resulting fiber is preserved. The fibers are dense, without any detectable porosity
even under very high resolution (<1 nm) transmission electron microscopy
(TEM), and show a very smooth surface. The considerable mass loss during
thermal treatment causes a shrinkage of the fiber diameter by 30-40%. Usually
this is regarded as unfavorable, but this shrinkage can actually be an advantage
for the production of ceramic fibers with extremely small diameters [113, 114].
6
Chemical and Physical Properties of Si/B/N/C Ceramics
6.1
During their genesis all precursor derived ceramics pass through an

amorphous state which in certain cases is stable up to the respective
decomposition temperature. The propensity to crystallize is strongest for the
binary silicon and boron nitrides and carbides, while in particular quaternary
materials out of the Si/B/N/C system show the strongest resistance towards
crystallization. It is interesting to note that in any case investigated, so far, the
amorphous multinary ceramics decompose into the binary border phases
upon crystallization - crystalline ternary silicon boron nitrides or carbides are
not known to date; cf. [9].
When trying to characterize such solids that lack translational symmetry,
one has to overcome some awkward problems known from the chemistry and
physics of glasses. Presumably, their structures correspond to some random
covalent network. However, no analytical tools are available that would allow a
conclusive and complete determination of structures, which constitute the
basis for any deeper understanding of the properties of these materials. Clearly,
as a first step the chemical composition has to be determined and, moreover,
the homogeneity of the element distribution has to be checked over different
length scales. To gain further reliable insights into the structural features, as
many different structure sensitive probes as possible have to be employed. The
best way to supplement these disjunct pieces of information to obtain a full
microscopic structural model appears to be the use of computational modeling

techniques. However, regarding the characterization of bulk materials, there
are still some deficiencies because of macroscopic pieces lacking.
6.2
Chemical Analyses
As precise as possible a knowledge of the chemical composition is a crucial

pre-requisite to any serious attempt to determine the geometric and
topological structure of a multi-component random network. However,
chemical analyses of refractive materials like the silicon boron carbonitrides
discussed here is a difficult task. In general, there exist two different classes of
procedures, destructive ones, such as wet chemical digestion, and
non-destructive (physical) ones.
Unfortunately, most of the physical methods exhibit low penetration
depths, and thus are basically surface sensitive. More importantly, the results
obtained strongly depend on the matrix in which the elements of interest are
embedded. This leads to significant systematic errors as long as no suitable
references with a known composition, as close as possible to the analyte, are
available. Methods based on the atomic number, such as the wavelength
dispersive X-ray microanalysis (WDX) [115] and the energy dispersive
X-ray microanalysis (EDX) [116], also suffer from substantial errors due to
absorption of the soft X-rays of the relatively light elements (B, C, N, O). In
principle, the restrictions mentioned apply to all physical tools for solid state
analysis such as Rutherford-backscattering-analysis (RBS) [117], electron
energy loss spectroscopy (EELS) [118], X-ray fluorescence spectroscopy
(XRF), and other related methods. However, these techniques have significant
advantages over destructive techniques with respect to the determination of
relative concentrations and spatially varying compositions. The best lateral
resolution presently achievable is 0.8-1.0 nm (EELS with a field emission
electron source). Furthermore, in certain contexts it might be of interest to
recover the sample undamaged.
When trying to determine the 'true' chemical composition of a sample, the
elements one focuses on have to be brought into a state virtually free of
interactions. This is achieved by wet chemical analyses which include as a first
step dissolution of the solid analyte. In the case of the extremely refractive
carbide nitride ceramics the digestion has to proceed employing rather harsh
conditions. The treatment under pressure, sometimes assisted by microwave
irradiation, using a mixture of HF/HNO3/H2SO4(HC1) represents a convenient
process for the digestion of a ceramic [119, 120]. Alternatively, the "alkali
fusion" in a melt of Na2CO3/K2CO3 serves as another convenient digestion
procedure suitable for Si/B/N/C ceramics. From such solutions boron and
silicon can be determined simultaneously by means of ICP-OES. The
quantitative analysis of carbon, nitrogen, and oxygen is usually carried out
in a hot gas extraction analyser, determining the latter two elements
simultaneously. At first glance, digestion with elemental fluorine and
quantitative detection of the resulting volatile fluorides by infrared spectros-
copy appears to be an attractive approach [121], which, however, is still

waiting for a thorough validation.
A survey of the most common analytical procedures as applied to "SiBN3C"
shows that the so-called destructive methods, especially ICP-OES and hot gas
extraction, provide much more reliable results than the non-destructive
physical techniques [25]. It should be noticed that no digestion procedure
allows for a determination of all elements of interest (Si, B, N, C, 0)
simultaneously.
6.3
Structure Determination and Structural Features
The geometric structure of a given amorphous material is characterized by its

short range (SRO), medium range (MRO), and long range order (LRO). This
kind of classification is reasonable since different analytical tools are needed
for each range, and the complexity of the task and the ambiguity of the results,
increase with the length scale. For obtaining information about SRO a large
number of probes is available. Besides spectroscopic methods (IR, Raman,
NMR, XANES, EELS), techniques based on scattering processes (neutron,
X-ray and electron diffraction, EXAFS, XELNES) can be applied. The choice is
much more limited for medium and long range structural features, which can
be extracted from pair correlation functions, determined by X-ray, neutron, or
electron diffraction, and, in particular, from advanced NMR experiments.
6.3.1
Short Range Order
From conventional 29Si- and n B-magic angle spinning (MAS) NMR spectra the
local structural units, i.e., the polyhedra associated with silicon and boron,
have been determined [122, 123]. The spectra for the ceramics derived from
TADB, TSDE, and MADB (Eqs. 1 and 6) are characterized by a remarkable
similarity among each other. By comparing the 29Si chemical shifts from the
amorphous compounds with the shifts from crystalline /?-Si3N4, one concludes
that silicon exhibits a fourfold tetrahedral coordination by nitrogen. The
lineshapes of the H B MAS spectra exhibit quadrupolar asymmetry parameters
near 0.1 (coupling constant 2.9 MHz) and isotropic chemical shift values of
30 ppm, from which a trigonal planar coordination of boron with three
nitrogen atoms can be deduced. While the n B MAS signals exactly match those
of the references [124] the 29Si lines show significantly larger half widths. This
has been traced back to an inhomogeneous linebroadening caused by the
distortion of the local environment of silicon. Since the spread in Si-N
distances, according to the pair distribution function (PDF) curves, is
remarkably small, the major contribute should come from the angular
distortion. The role of carbon in the quaternary networks discussed here is still
under investigation, but apparently anionic carbon, i.e. carbon directly bonded
to silicon or boron, can be excluded in most of the ceramics. Instead, sp 2
hybridized carbon seems to be present in two different substructures: first,
incorporated within the cationic positions of the random network, replacing

boron or silicon and, second, located well apart from the network in an
environment which still has to be determined [122]. Investigations of electron
spectroscopy for chemical analysis (ESCA) have confirmed the presence of
C—N bonds [20].
15
N MAS spectra of 15N enriched samples of SiBN3C reveal two signals with
isotropic chemical shifts at 45 and 58 ppm, respectively. These shifts lie well in
the region of the chemical shifts of hexagonal BN and Si3N4 (a- and /?-), and
thus are indicative for nitrogen in a trigonal planar coordination by boron and
silicon [125].
The NMR-results are in good agreement with the findings from X-ray and
neutron scattering experiments on the amorphous ternary nitride Si3B3N7, a
compound prepared from the single source precursor TADB and ammonia
[126, 127]. The maxima in the pair correlation functions of this nitride show
distinct peaks at approximately 1.43 and 1.75 A, which can clearly be
attributed to B-N and Si-N distances. These experimental data also provide the
coordination numbers for the first sphere of silicon and boron, indicating a
mean coordination number of 2.95 for boron (almost trigonal planar BN3
groups) and a mean coordination number of 3.8 for silicon (slightly distorted
SiN4 tetrahedra).
The PDFs of Si3B3N7 and SiBN3C as obtained through electron diffraction
confirm the presence of BN3 and SiN4 units [128] and also show that similar
short range order is present in these two networks, although the PDF peaks of
SiBN3C are somewhat broadened.
6.3.2
Medium and Long Range Order
The most powerful tool to study the intermediate range order in amorphous
Si/B/N/(C) networks is MAS-NMR spectroscopy including advanced double
resonance techniques such as n B-{ 29 Si}rotational echo double resonance
(REDOR), 29 Si-{ n B} rotational echo adiabatic passage double resonance
(REAPDOR), and 13 C-{ n B} REAPDOR. These methods utilize the heteronu-
clear coupling between the observed nucleus and a nucleus in the second
coordination sphere to get information about the internuclear distance of the
two nuclei.
In nitrides, the second coordination sphere of silicon consists of eight
cations, whereas in the second coordination sphere of boron a maximum of six
cationic species is possible. In the case of a fully random cationic partial
structure, these positions should be equally occupied by silicon and boron.
However, the experimental results are in striking contrast to that expectation:
in Si3B3N7, silicon is surrounded by 1.8 boron and 6.2 silicon atoms, while
boron has only 1.4 silicon, plus 4.6 boron, neighbors in the second
coordination sphere. Accordingly, the medium range order of the investigated
random network Si3B3N7 is characterized by an agglomeration of B-N-B
linkages and Si-N-Si linkages to form boron-rich and silicon-rich domains
(Fig. 8). Similar results have been obtained for SiBN3C [122].
B
\
Si N—Si N
\ / ol i y"
N—B :N—SI-N^
Si N—Si Si' Si
/ N
B ' ' ~"Si
Si
a b
Fig. 8a, b. Second coordination spheres in Si3B3N7 for: a boron; b silicon
For Si3B3N7 the maximum in the PDFs (X-ray and neutron) at about 2.5 A
can be assigned to either an N-N or B-B separation in a BN2 or NB2 triangle.
The next maximum at ca. 2.8 A represents two adjacent nitrogen atoms
coordinated to silicon but may also include a Si-B interatomic distance. With
neutron contrast techniques applied to nitrogen, additional information
concerning the higher coordination shells has been gained. Regarding the
third coordination shell, four possible chain configurations of Si-N and B-N
pairs connected via three bonds are possible. The dihedral angle distribution
for each chain has been extracted snowing all possible combinations of the
precursor unit B/Si-N-B/Si-N present in this nitride ceramic [126, 127].
Finally, the spatial homogeneities of SiBN3C and Si3B3N7 have been checked
with high resolution mapping of the elements Si, B, N, and C through electron
spectroscopic imaging (ESI). It turns out that both compounds, Si3B3N7 and
SiBN3C, are highly homogeneous above a lateral resolution of the method
applied in the range of 0.8-1.0 nm [128].
6.3.3
Modeling of Si/B/N Random Networks
Complementing the experimental techniques discussed in the previous

sections, models for Si/B/N networks have been constructed [129]. Since
simulating the actual synthesis route via the sol-gel process and pyrolysis with
sufficient accuracy is still too time-consuming, a large number of candidate
models were generated using alternative routes, e.g., freezing from the melt via
MD-simulations, random-close-packing of anions and cations (RCP), molec-
ular modeling (MM), and construction from crystal fragments with a diameter
of up to 0.7 nm (Fig. 9). For the static RCP- and MM-models, the information
that Si and B prefer to be fourfold and threefold coordinated by N,
respectively, served as input. Relaxations of these candidates with the help
of specially fitted empirical potentials [130] produced the final models. The
experimentally observed short-range order is correctly reproduced by all
models but, interestingly enough, the best agreement with the experimental
neutron and X-ray pair correlation functions is found for the heterogeneous
fragment model (Fig. 10). Similarly, the medium range order information
gained from NMR-experiments suggests that the more heterogeneous model is
to be preferred. For all models, the computed density (p = 2.2-2.8 g cnT 3 )
exceeds the one found experimentally (p = 1.8 g cm" 3 ), but again the
fragment model (p = 2.2 g cm" 3 ) comes closest. Clearly, the degree of
Fig. 9. Structure model for Si3B3N7: Si (black), B (light gray), N (white, small). The
heterogeneous cation distribution is clearly visible
heterogeneity on the scale below 1 nm is quite high, and presumably related to

the specific details of the synthesis route.
6.4
Bulk Properties of Polyborosilazane Derived Ceramics
The structures as well as the bulk properties of the quaternary Si/B/N/C

ceramics strongly depend on details of their synthesis route. Most influential
are the choice of the molecular precursor and of the linking agent, the
conditions maintained during the polymer-to-ceramic conversion, and the
possible exposure to oxygen during the process. As a consequence, experi-
mental data about any property only apply to a specific sample. Thus, these
data are virtually meaningless without the full details of the sample's history
being given.
With respect to the intended applications, the thermal durability at inert
conditions and, in particular, in air is the most crucial feature. Regrettably,
often this aspect is discussed in the literature in a confusing manner. Mostly,
the onset of mass loss at raising the temperature as determined by Thermo-
gravimetric Analysis (TGA), which is a dynamic method, is taken as the limit of
structural stability. Structural transformations, usually occurring at lower
temperatures, as well as long term stability and the behavior of the material in
atmospheres relevant for applications (e.g. air) are frequently neglected.
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6

R[A]
Fig. 10. Comparison between experimental (dashed line) and simulated (solid line) pair
correlation functions (PCF) for neutron and X-ray scattering for the optimized crystal
fragments model
Figure 11 shows TGA measurements for some Si/B/N/C ceramics which

have been synthesized through polycondensation of the corresponding single
source precursor and miscellaneous crosshnking reagents. The onsets of
weight losses for the investigated compounds are included in Table 2.
Si/B/N/C ceramics derived from single source precursors DSMB, TSMB, and
TSDE, utilizing methylamine for crosshnking, show the best thermal stability
as denned by the onset of weight loss in an inert atmosphere. In all these cases,
only slight evolution of nitrogen starts at ~1800 °C, and even at 2000 °C the
mass loss has not exceeded 20%. Using other crosshnking reagents instead of
methylamine (cf. Sect. 3.3.1), polymerization of TABB (dimethylamide deriv-
ative of TADB) and subsequent pyrolysis results in Si/B/N/C ceramics with less
thermal stability already decomposing at temperatures between 1358 and
1812 °C (Fig. 12) [87].
It should be noticed that the structural stability of an amorphous network
must not be characterized by the numerical value of the onset of weight loss
only, or by the maximum temperature the ceramic can temporarily be exposed
to. Since structural rearrangements are kinetically controlled and may
continue over a long period of time, even at high temperatures, long term
TGA measurements have to be performed at determining the real thermal
durability of a ceramic. This has been done in the case of the amorphous
network SiBN3C pyrolyzed at 1500 °C which, after 20 h at 1700 °C in pure
TG1%
800 1000 1200 1800 2000
temperature / °C
Fig. 11. High temperature TGA curves of ceramics derived from single source precursors
and methylamine as crosslinking agent
800 1000 1200
temperature / DC
Fig. 12. High temperature TGA curves of ceramics derived from the single source precursor
TABB and different crosslinking agents
nitrogen, showed only an initial mass loss of 3% without any crystallization of

the amorphous network taking place. Depending on composition, way of
synthesis, and especially on the oxygen content, Si/B/N/C ceramics start to
crystallize between 1500 °C and 1900 °C into composites consisting of a/jS-SiC
and a//?-Si3N4. Remarkably, according to XRD measurements formation of
binary crystalline boron nitrides or carbides has never been observed during
decomposition of this quaternary system [20, 91].
In some Si/B/N/C ceramics silicon nitride is still present even after
treatment at temperatures beyond 2000 °C, although silicon nitride normally
decomposes at 1600 °C [91]. Moreover, CALPHAD (calculation of phase

diagrams [131]) calculations show that in the presence of free carbon Si3N4
should react at 1484 °C to form SiC and nitrogen. A possible explanation for
this behavior has recently been presented by Aldinger and coworkers [132,
133] who observed turbostratic BN(C) segregations along grain boundaries of
nanosized silicon carbide and silicon nitride crystals in a crystalline composite
ceramic made from a borane modified polyhydridomethylsilazane (Fig. 13).
The encapsulation of nanocrystalline Si3N4 grains by the BN(C) phase is
supposed to have two effects. First, BN may act as a diffusion barrier and in
addition binds free carbon, thus decreasing and inhibiting the reaction of C
with Si3N4. Furthermore, the internal nitrogen pressure is probably increased
due to the adjacent BN(C) phase leading to a higher decomposition
temperature of silicon nitride (at a nitrogen pressure of 1 MPa the decom-
position temperature of Si3N4 is as high as 2034 °C).
Compared to BN, Si3N4 or other non-oxide ceramics, the oxidation
resistance of Si/B/N/C ceramics is amazingly high. When ceramics derived
•I M l
Fig. 13. TEM micrograph of a crystalline Si/B/N/C ceramic composite [132]

from, e.g., TADB, TSDE, TSDM, or TSMB were exposed to pure oxygen at
temperatures up to 1400 °C during a TGA-measurement, an increase in weight
of only 2-3% was observed, without any degradation or crystallization taking
place (Fig. 14).
The oxidation behavior of SiBN3C in air has been investigated over a longer
period of time employing isothermal TGA measurements at temperatures
between 1520 and 1550 °C. The recorded weight loss (evaporation of B2O3)
reveal a parabolic gradient (Vt-law), and the overall oxygen content ap-
proaches a low limit value indicating a diffusion-controlled oxidation process
that is strongly impeded kinetically. At the beginning of this oxidation process
a protective double layer on the surface of this ceramic material forms, leading
to an excellent oxidation resistance even at high temperatures: the outer layer,
rich in Si and O but depleted with respect to B, acts as a diffusion barrier for
oxygen, while the inner layer, rich in N and B but with less Si, suppresses
cation diffusion.
Depth profile investigations by Secondary Neutral Mass Spectrometry
(SNMS) of SiBN3C powder, oxidized at 1520 °C in air, together with a SEM
image of the intersection of a SiBN3C fiber also exposed to air (50 h at
1500 °C) clearly show the presence of that double layer of approximately
1.2-1.4 um thickness (Fig. 15).
7
SiBN3C-Fibers
7.1
Technical Performance
While the processing of binary ceramics to fibers has been intensively

investigated, and is in some cases well established on a technical scale, only a
TG /%
10-
TSDM + MeNH,
-20
200 400 600 800 1000 1200 1400
temperature / °C
Fig. 14. TGA curves of some selected Si/B/N/C ceramics in pure oxygen atmosphere
bO
0
Al:
40 ywJK ,N
/Si ^ ^ ^ l ^ ^
30-
I \J
20-
C/% c
10- B
1000 2000 3000 4000 , 5000 6000

f/s
Fig. 15a, b. Formation of a double layer on SiBN3C after heat treatment in air: a SNMS
(secondary neutral mass spectra) of the surface of SiBN3C powder, oxidized at 1500 °C in air
(erosion rate: 2 |im in 3000s); b scanning electron micrograph of an oxidized SiBN3C fiber
(1500 °C, 50 h in air)
few attempts [114, 134] to spin fibers of multinary systems, not to mention
amorphous ones, have been reported. Thus the SiBN3C fiber is the first
representative of a new, very promising family of high performance ceramic
fibers [103, 114]. The fibers have been produced from the meltable
JS[-methylpolyborosilazane as a multinlament of up to 300 individual fibers
with adjustable diameters of 8-15 urn and a length exceeding 200 m. The
fibers have a very smooth surface and very constant diameter. Their oxygen
content is below 1 wt%. In general, the fibers reflect the properties of the bulk
SiBN3C ceramic, and show an outstanding thermal stability, a high resistance
against crystallization, and, in particular, oxidation. The amorphous state is
retained at least up to 1800 °C, and is not subject to microstructural changes
up to this temperature. This results in a remarkably good high-temperature
creep resistance. Under oxidizing conditions the fibers can withstand

temperatures of up to 1500 °C for long periods while retaining almost all of
their mechanical strength. This superior oxidation stability is due to the
formation of a protective double layer, upon exposing this material to air at
1400 °C (cf. Sect. 6.4). Thus, the maximum operating temperature for SiBN3C
in air is approaching 1500 °C, which is significantly higher than that for all
competing non-oxidic materials systems reported so far. Even after temper-
ature cycling in air (from 20-1500 °C, 5 cycles of 10 h each), the fibers show a
crack-free oxidation layer with good adhesion to the fiber core, and the bulk
material as well as the oxide layer remain amorphous. The tensile strength
of the oxidized fiber is as high as that of the unoxidized one. Another
advantageous feature of the fiber is the high degree of resistance to corrosion
by liquid metals, such as copper and silicon, most likely due to the existence of
the protective double layer and the stability of the random network. The
corrosion resistance to liquid silicon enables the use of the ceramic fibers for
reinforcement of ceramic matrix composites (CMCs) employing the low-cost
method of liquid silicon infiltration [135]. Another important aspect for the
production of CMC materials is related to the bonding strength between the
fiber and the surrounding matrix material: it must be weak enough to ensure
that cracks will be deflected around the fiber and stopped, and at the same
time it must be strong enough to ensure that the fibers cannot be easily pulled
out. The current method for achieving the right bonding strength consists in
coating the fiber before embedding it in the matrix. With SiBN3C, this costly
additional step seems to be avoidable since the double-layer formed during an
oxidative curing step appears to provide the proper fiber-matrix bonding
requirements. This curing step is easily performed in a continuous process at
1500 °C in air, and can directly follow the pyrolysis and sintering steps.
In conclusion, the superior properties of SiBN3C fibers (cf. Table 4) can
be summarized as follows: excellent resistance to oxidation, suitable for
application in air up to 1500 °C, resistance to molten silicon at 1600 °C,
suitable for application up to 1800 °C in an inert atmosphere, no microstruc-
tural changes up to 1800 °C, low specific weight (1.8 g cm" 3 ), small coefficient
of thermal expansion (3 x 10~6 K"1), and a thermal conductivity of approx-
imately 3 W m" 1 K"1 at 1200 °C [114].
7.2
Comparative Assessment of SiBN3C- and SiC-Fibers
Until now, only a handful of ceramic fibers for use in CMCs, mainly fibers
made from silicon carbide, have reached the market. While Ube Industries,
Nippon Carbon, and, lately, Dow Corning produce SiC fibers, 3 M has
developed some oxidic ceramic fibers (e.g. aluminoborosilicate, aluminosilica,
or alumina) for this purpose. Since up to now for high-temperature
applications only fibers made from SiC or SiBN3C are suitable, the most
important features of these two classes will now be discussed.
In contrast to the different polycrystalline SiC fibers, SiBN3C fibers exhibit a
higher tensile strength at elevated temperatures combined with an excellent
Table 4. Mechanical data of SiBN3C-fibers compared with the requirements of potential users (European turbine producers) and to various SiC-fiber
types
Properties End user SiBN3C [137] Sylramic [138-140] HiNicalon [138-142] Ube SA [140, 143]
specification (Dow corning) (Nippon carbon) (Ube industries)
Oxygen content (% in weight) _ 1.0 0.2 0.8 0.3

Microstructure - Amorphous Crystalline Partial crystalline Crystalline
Tensile strength
at RT (GPa) 3.0 4.0 2.8 3.0 2.5
at 1400 °C in Ar (% RT-value) 83 (1500 °C) 95 95 70 95
Strength (% of RT value) after exposure to air
at 1300 °C for 100 h - - - 23 55
at 1500 °C for 50 h - 80 - 0 _
E-modulus
at RT (GPa) 300 290 420 280 300
at 1400 °C in Ar (% RT-value) 83 90 - 70 -
Strain to failure 1.0 1.0 0.8 1.3 0.7
Creep parametera >0.4 0.8 0.4 0.1 0.7
Subcritical crack growth - No Yes No Yes
Coefficient of thermal expansion (BT1) 3.5 x 10"6 3.5 x 10"6 4.0 x 10~6 4.0 x 10~6 4.0 x 10~6
Specific weight (g cm"3) <5.0 1.85 3.1 2.7 3.1
Thermal conductivity (W m"1 K"1) - >3 46 5 64
Resistance to molten silicon - Yes No No No
Diameter (um) 10-150 12-14 10 14 10
1
Bend-stress-relaxation measurements at 1400 °C.
creep resistance at high temperatures, even up to 1800 °C in nonoxidizing

atmospheres, and up to 1500 °C under oxidizing conditions, for long periods,
while retaining almost all of their mechanical strength. After exposure to air
(15 h at 1500 °C) the HiNicalon SiC fiber is largely oxidized, losing nearly all
of its mechanical strength, whereas the SiBN3C fiber shows a crack-free oxide
layer with good adhesion to the fiber core, the tensile strength of the fiber
remaining nearly as high as that of an unoxidized sample. Even after
temperature cycling, the SiBN3C fiber remains amorphous both in the bulk
material and in the oxide layer. The existence of this protective double layer on
the SiBN3C fiber and the stability of the random network are the most likely
reasons for the high degree of resistance to corrosion by liquid metals, such as
copper or silicon. This makes SiBN3C suitable for convenient low-cost
production techniques for CMCs such as liquid silicon infiltration (LSI). In
contrast, SiC fibers tend to recrystallize in such media leading to grain growth,
culminating in a significant loss of strength. From the point of view of
processing technology, this latter fact prevents the use of the well established
LSI method for the production of CMCs based on SiC fibers.
The thermomechanical data of SiBN3C fibers available up to now, on
laboratory scale samples, are summarized in Table 4, along with the
corresponding data for various SiC fibers. Regarding tensile strength and
mechanical strength as measured by Young's modulus, all ceramic fibers
exhibit comparable values at ambient temperature. However, the outstanding
oxidation resistance and creep behavior of the amorphous SiBN3C fibers lead
to excellent thermomechanical data in air and at high temperatures.
The superior performance of SiBN3C fibers compared with crystalline SiC
fibers is not only due to the spatial homogeneity of the polymeric intermediate
PBS-Me down to the atomic level, which is a crucial prerequisite for the
formation of an amorphous and flaw-free fiber. The protective double layer
also has to be mentioned, which becomes an integrated part of the fiber during
processing and is largely responsible for the high oxidation resistance of this
material. The major drawback of the SiC fibers is their crystalline nature,
which significantly limits the creep resistance and makes the fibers intrinsi-
cally prone to subcritical crack growth leading to a considerably reduced
lifetime at temperatures exceeding 1300 °C, especially in air.
8
Concluding Remarks
Thinking conventionally, the approach of employing amorphous solids as
high-performance ceramics would imply a contradiction in terms since
such materials are thermodynamically unstable with respect to transformation
or decomposition to crystalline phases, which would exclude them from
applications in safety sensitive areas at high temperatures. However, in
constructing devices, e.g., engines or turbines, usually different materials are
combined, and the whole assembly is by no means in thermodynamic
equilibrium. So, it is the kinetic stability that counts in the first place, though,
incontestably, high thermodynamic stability represents an attribute highly
favorable for materials intended for an application at extremely demanding

conditions.
As compared to the expectations based on thermodynamic considerations,
the excellent overall performance of the amorphous Si/B/N/C-ceramics is a real
surprise. The main reason for their high kinetic stability can be traced back to
the fact that the underlying random network is constituted of elements which
form strong covalent bonds among each other and are bounded to at least
three neighboring atoms. Structural rearrangements within such networks
would require at minimum two bonds to be cleaved at the same atom and at
the same time, which only occurs with low probability. Furthermore, the
thermodynamic driving force towards crystallization is minimized since the by
far prevailing contribution to the heat of formation of compounds in the
system Si/B/N/C and its subsystems results from local, covalent bonds, and the
energetic benefit from long range order is negligibly small. This view gains
support from the observation that at introducing more polar or more metallic
bonding contributions the materials become more inclined to phase separa-
tion and crystallization. Instructive examples are polymer-derived Si/Ti/N [53]
or Si/Al/N [50] materials which show a strong tendency to form crystalline
composites at pyrolysis.
These features can account for the high durability with respect to weight
loss and to microstructural transformations, while the unforeseen resistance
against oxidation has to be traced back to the formation of a protective double
layer. The outer layer, mainly consisting of silicon, oxygen, and nitrogen, acts
as a diffusion barrier for oxygen, while the inner, enriched with respect to
boron and slightly depleted in silicon, is a barrier against cation diffusion.
The bulk properties discussed thus far strongly depend on how the
respective preceramic polymer has been prepared. The main issues are design
of the molecular precursors (single source or multiple source) and choice
of the linking agents. As a rule, co-condensation of mixtures of precursor
molecules leads to spatial inhomogeneities in the polymer which induces
crystallization and phase separation during pyrolysis. Bridges between the
cations consisting of more than one nitrogen or carbon atom obviously
constitute weak links, as can be seen from guanidinium- or cyanamide-
containing polymers that yield ceramics of disappointing low thermal stability.
From polymeric intermediates involving C2-units, as formed during hydrob-
orations, ceramics richer in carbon are obtained. These show a high stability
against weight loss, although they start to crystallize by 1700 °C [134].
Particularly well performing ceramics, with respect to the combined relevant
properties, seem to be accessible when using small single source precursor
molecules of the general constitution (CH3)xCl(3^x)Si-Y-BCl2 (Y = CR2 or
NR with R = H or CH3; x = 0, 1, 2), and methylamine as a linking agent.
Amazingly, materials made from TADB, the single source precursor that
has led to the discovery of the extremely promising properties of Si/B/N/C
amorphous networks [17,136], still belong among the top performing systems.
This seems to be related to the fact that Si—N and B—N bonds are reinforced
by ^-interactions which are effective in a synergetic manner within the Si-N-B
networks.
The production of the polymer has reached pilot plant scale (batches of
150 kg), and the polymer can be spun and pyrolyzed continuously to long
fibers. However, the mechanical properties of fibers made in such large
quantities still suffer from too high an oxygen contamination and lag behind
by about 20% with respect to tensile strength, as compared to laboratory scale
products.
TADB-derived ceramic fibers, with the idealized composition SiBN3C, do
not reach the E-modulus of the most advanced SiC fibers at room temperature.
However, they are clearly superior to the latter in a crucial point, namely the
drop of the E-modulus and of the creep resistance at high temperatures. SiC
fibers already lose a large part of their mechanical strength below 1400 °C, as
can be measured by creep resistance. These limitations are fundamental in
nature, since they are related to grain boundary sliding and thus to the
crystallinity of SiC. In contrast, amorphous SiBN3C fibers do not show any
grain boundaries and, moreover, the concentration of microstructural flaws is
extremely low.
The prospects of widespread application of amorphous Si/B/N/C ceramics
are rather promising. These expectations are not only based on the attractive
properties (low density, low thermal coefficient of expansion, low thermal
conductivity, high thermal shock resistivity, and good mechanical perfor-
mance, in particular at high temperatures) but also on the broad choice in
ways of processing. Many of the precursor syntheses as worked out in the
laboratory have the potential for being scaled up to technical dimensions,
which has been already demonstrated for the TADB-based route.
9
References
1. Hardtl KH (1976) Ferroelectrics 12: 9
2. Jaffe B, Cook WR, Jaffe H (1971) In: Roberts JP, Popper P (eds) Piezoelectric Ceramics.
Academic Press, New York
3. Wu MK, Ashburn JR, Torng CJ, Hor PH, Meng RL, Gao L, Huang ZJ, Wang YQ, Chu
CW (1987) Phys Rev Lett 58: 908
4. Simon A, Kohler J, Borrmann H, Gegenheimer B, Kremer R (1988) J Solid State Chem
77: 200
5. Cava RJ (2000) J Am Ceram Soc 83: 5
6. Kunz KP, Sarin VK, Davis RF, Bryan SR (1988) Mat Sci Eng A105: 47
7. Kijima K, Shirasaki S (1976) J Chem Phys 65: 2668
8. Landolt-Bornstein (1997) Numerical data and functional relationships in science and
technology. Springer, Berlin, Heidelberg, New York
9. Aldinger F, Seifert HJ (2001) Struct Bond (this volume)
10. Chantrell PG, Popper EP (1964) In: Popper EP (ed) Special Ceramics. Academic Press,
New York, p 87
11. Winter G, Verbeek W, Mansmann M (1975) Patent DE3892583; (1974) Chem Abstr 81:
126,134
12. Lange H, Wotting G, Winter G (1991) Angew Chem 103: 1606; (1991) Angew Chem Int
Ed Engl 30: 1579
13. Yajima S, Hayashi J, Omori M (1975) Chem Lett 931
14. Yajima S, Hayashi J, Omori M, Okamura K (1976) Nature 261: 683
15. Seyferth D, Wisemann GH (1984) Ultrastruc Process Ceram Glasses Compos Proc Int
Conf
16. Seyferth D, Wisemann GH (1984) Polym Prepr Am Chem Soc Div Polym Chem 25: 10
17. Baldus HP, Wagner O, Jansen M (1992) Mat Res Soc Symp Proc 271: 821; (1992) Pat
EP0502399A2
18. Jansen M, Jungermann H (1997) Curr Op Solid State Mater Sci 2: 150
19. Jansen M (1997) Sol State Ionics 101/103: 1
20. Baldus HP, Jansen M (1997) Angew Chem 109: 338; (1997) Angew Chem Int Ed Engl 36:
328
21. Baldus HP, Wagner O, Jansen M (1994) Key Eng Mater 89: 75
22. Miiller U (2000) PhD thesis, University of Bonn, Germany; (2000) Pat DE10045427.5
(publication in March 2002)
23. Jansen M, Jaschke T (1999) Z Anorg Allg Chem 625: 1957
24. Gasparis-Ebeling T, Noth H (1990) Chem Ber 123: 261
25. Kroschel M (2001) PhD thesis, University of Bonn, Germany
26. Kroschel M, Jansen M (2000) Pat DE10045428.3 (publication in March 2002)
27. Wannagat U, Moretto H, Schmidt P, Schulze M (1971) Z Anorg Allg Chem 381: 288
28. Soundararajan R, Matteson DS (1990) J Org Chem 55: 2274
29. Soundararajan R, Matteson DS (1995) Organomet 14: 4157
30. Jungermann H, Jansen M (1998) Pat EP0973781A1
31. Jungermann H, Jansen M (1999) Mat Res Innovat 2: 200
32. Riccitiello SR, Hsu MTS, Chen TS (1991) Pat US4987201
33. Weinmann M, Haug R, Bill J, Aldinger F, Schuhmacher J, Mttller K (1997) J Organomet
Chem 541: 345
34. Ruwisch LM, Diirichen P, Riedel R (2000) Polyhedron 19: 323
35. Jones PR, Myers JK (1972) J Organomet Chem 34: C9
36. Weinmann M, Kamphowe TW, Fischer P, Aldinger F (1999) J Organomet Chem
592: 115
37. Smith K (1974) Chem Soc Rev 3: 443
38. Gerrard W, Lappert MF (1958) 58: 1081
39. Jaschke T, Jansen M (2000) Pat DE10045050.4 (publication in March 2002)
40. Jaschke T, Jansen M (2000) Pat DE10108069.7 (publication in March 2002)
41. Jaschke T, Jansen M (2001) in preparation
42. Ruwisch LM, Riedel R, Klingebiel U, Noltemeyer M (1999) Z Naturforsch B 54: 624
43. Gastreich M, Marian CH, Jungermann H, Jansen M (1999) Eur J Inorg Chem 1: 75
44. Miihlhauser M, Gastreich M, Marian CH, Jungermann H, Jansen M (1996) J Phys Chem
100: 16,551
45. Becke AD (1993) J Chem Phys 98: 5648
46. Lee C, Yang W, Parr RG (1988) Phys Rev B 37: 785
47. Miehlich B, Savin A, Stoll H, Preus H (1989) Chem Phys Lett 157: 200
48. Jaschke T (1997) Diploma thesis, University of Bonn, Germany
49. Dietz JW (1969) Patent US3,676,343
50. Loffelholz J, Jansen M (1995) Adv Mater 7: 289
51. Perchenek N, Baldus HP, Loffelholz J, Jansen M (1994) Pat EP0623643A2
52. Loffelholz J (1994) PhD thesis, University of Bonn, Germany
53. Loffelholz J, Engering J, Jansen M (2000) Z Anorg Allg Chem 626: 963
54. Schmidt R, Sporn D, Schubert U (1996) Ceram Forum Int 73: 182
55. Hsu MTS, Chen TS, Riccitiello SR (1991) J Appl Polym Sci 42: 851
56. Riccitiello SR, Hsu MTS, Chen TS (1988) Patent US4,767,728
57. Hsu MTS, Chen TS, Riccitiello SR (1986) Polym Prepr Am Chem Soc, Div Polym Chem
27: 261
58. Adrian T, van Aefferden B, Habel W, Mayer L, Nover C, Sartori P (1992) Patent
EP497,171
59. Von Gutenberger P, Habel W, Nover C, Sartori P (1993) J Organomet Chem 453: 1
60. Riccitiello SR, Hsu MT, Chen TS (1991) Pat US4,987,201
61. Riedel R, Kienzle A, Szabo V, Mayer J (1993) J Mater Sci 28: 3931
62. Shiina K, Kumada M (1958) J Org Chem 23: 139
63. Babonneau F, Maquet J, Bonhomme C, Richter R, Roewer G, Bahloul D (1996) Chem
Mater 8: 1415
64. Hemida AT, Pailler R, Birot M, Pillot JP, Dunogues J (1997) J Mater Sci 32: 3237
65. Takamizawa M, Kobayashi T, Hayashida A, Takeda Y (1984) Patent US4,550,151
66. Takamizawa M, Kobayashi T, Hayashida A, Takeda Y (1984) Pat US4,604,367
67. Noth H (1961) Z Naturforsch B Anorg Chem Org Chem 16: 618
68. Seyferth D, Plenio H (1990) J Am Ceram Soc 73: 2131
69. Seyferth D, Plenio H (1989) Pat US5,171,736
70. Zank GA (1991) Patent US5,169,908
71. Su K, Remsen EE, Zank GA, Sneddon LG (1993) Chem Mater 5: 547
72. Wideman T, Cortez E, Remsen EE, Zank GA, Carroll PJ, Sneddon LG (1997) Chem
Mater 9: 2218
73. Wideman T, Fazen PJ, Su K, Remsen EE, Zank GA, Sneddon LG (1998) Appl
Organomet Chem 12: 681
74. Aldinger F, Weinmann M, Bill J (1998) Pure Appl Chem 70: 439
75. Weinmann M, Aldinger F (1999) High temperature stable ceramics from inorganic
polymers. In: Bill J, Wakai F, Aldinger F (eds) Precursor-Derived Ceramics, Synthesis,
Structures and High Temperature Mechanical Properties. Wiley-VCH, p 83
76. Weinmann M, Haug R, Bill J, Aldinger F, Schuhmacher J, Miiller K (1997) J Organomet
Chem 541: 345
77. Weinmann M, Haug R, Bill J, de Guire M, Aldinger F (1998) Appl Organomet Chem 12:
725
78. Haug R, Weinmann M, Bill J, Aldinger F (1999) J Eur Ceram Soc 19: 1
79. Funayama 0, Arai M, Aoki H, Tashiro Y, Katahata T, Sato K, Isoda T, Suzuki T, Kohshi
I (1990) Pat EP404503A1
80. Funayama O, Arai M, Tashiro Y, Isoda T, Sato K (1991) Pat US5,030,744
81. Funayama O, Kato T, Tashiro Y, Isoda T (1993) J Am Ceram Soc 76: 717
82. Funayama 0, Nakahara H, Okoda M, Okumura M, Isoda T (1995) J Mater Sci 30: 410
83. Wideman T, Su K, Remsen EE, Zank GA, Sneddon LG (1995) Chem Mater 7: 2203
84. Wideman T, Su K, Remsen EE, Zank GA, Sneddon LG (1996) Mater Res Soc Symp Proc
410: 185
85. Jungermann H (1997) PhD thesis, University of Bonn, Germany
86. Jungermann H, Jansen M (1997) Pat EP819720A1
87. Jungermann H, Jansen M (1998) Materialwiss Werkst 29: 573
88. Srivastava D, Duesler EN, Paine RT (1998) Eur J Inorg Chem 6: 855
89. Riedel R, Kienzle A, Petzow G, Briick M, Vaahs T (1994) Pat DE4320782A1
90. Riedel R, Kienzle A, Dressier W, Ruwisch L, Bill J, Aldinger F (1996) Nature 382: 796
91. Weinmann M, Schuhmacher J, Kummer H, Prinz S, Peng J, Seifert HJ, Christ M, Miiller
K, Bill J, Aldinger F (2000) Chem Mater 12: 623
92. Jansen M, Kroschel M (2000) Z Anorg Allg Chem 626: 1634
93. Demmer S (2000) Diploma thesis, Technical College of Wiirzburg-Schweinfurt-
Aschaffenburg, Germany
94. Seyferth D, Wiseman GH (1984) Ultrastruc Process Ceram Glasses Compos Proc Int
Conf, p 265
95. Seyferth D, Wiseman GH (1984) Polym Prepr Am Chem Soc Div Polym Chem 25: 10
96. Narisawa M, Shimoda M, Okamura K, Sugimoto M, Seguchi T (1995) Bull Chem Soc
Jpn 68: 1098
97. Funayama O, Nakahara H, Tezuka A, Ishii T, Isoda I (1994) J Mater Sci 29: 2238
98. Seyferth D, Strohmann C, Dando NR, Perrotta AJ (1995) Chem Mater 7: 2058
99. Bahloul D, Pereira M, Goursat P (1993) J Am Ceram Soc 76: 1156
100. Han HN, Lindquist DA, Haggerty JS, Seyferth D (1992) Chem Mater 4: 705
101. Lewis RH, Maciel GE (1995) J Mater Sci 3: 5020
102. He JL, Scarlete M, Harrod JF (1995) J Am Ceram Soc 78: 3009
103. Baldus HP, Passing G (1994) Mat Res Soc Symp Proc 346: 617
104. Weisbarth R (2002) PhD thesis, University of Bonn, Germany
105. Liicke J, Baldus HP, Passing G, Rigby K, Vogt K (1998) cfi/Ber DKG 75: 18
106. Rooke MA, Sherwood PMA (1997) Chem Mater 9: 285
107. Haltrich M, Wahl G, Arndt J, Suchentrunk R (1997) Proc Electrochem Soc 97-25: 1223
108. Haltrich M (1999) PhD Thesis, University of Braunschweig, Germany
109. Ziegler G, Lucke J, Richter I, Suttor D (1997) Key Eng Mater 132/136: 1870
110. Zheng GB, Sano H, Uchiyama Y, Kobayashi K (1999) Key Eng Mater 159/160: 393]
111. Weinmann M, Kamphowe TW, Schuhmacher J, Miiller K, Aldinger F (2000) Chem
Mater 12: 2122
112. Yang H, Deschatelets P, Brittain ST, Whitesides GM (2001) Adv Mater 13: 54
113. Baldus HP, Perchenek N, Thierauf A, Herborn R, Sporn D (1997) Pat DE19530404A1
114. Baldus HP, Jansen M, Sporn D (1999) Science 285: 699
115. Giinzler G, Bock H (1990) Scanning electron microscopy. Springer, Berlin, Heidelberg,
New York
116. Russ JC (1984) Fundamentals of energy dispersive X-ray analysis. Butterworth, London
117. Diirr AC (1998) Diploma Thesis, University of Bonn, Germany
118. Hofer F, Warbichler P (1987) Microchim Acta 1: 125
119. Rings S, Sievers R, Jansen M (1999) Fresenius J Anal Chem 363: 165
120. Mann S, Geilenberg D, Broekaert JAC, Jansen M (1997) J Anal Atom Spectr 12: 975
121. Kipphardt H, Garten RPH, Jacob E, Broekaert JAC, Tolg G (1997) Microchim Acta 125:
101
122. van Wiillen L, Jansen M (2000) J Mater Chem 10: 1
123. van Wiillen L, Miiller U, Jansen M (2000) Angew Chem 112: 2574
124. Jeschke G, Kroschel M, Jansen M (1999) J Non-Cryst Solids 260: 216
125. Jeschke G, Jansen M (1998) Angew Chem 110: 1342
126. Hagenmayer RM, Miiller U, Benmore CJ, Neuefeind J, Jansen M (1999) J Mater Chem 9:
2865
127. Hagenmayer RM, Miiller U, Jansen M (2000) Physica B 276/278: 423
128. Heinemann D, Assenmacher W, Mader W, Kroschel M, Jansen M (1999) J Mater Res
14: 3746
129. a) Hannemann A, Schon JC, Oligschleger C, Jansen M (1999) Modelling of Si-B-N
ceramics, Proceedings of the workshop on "structure and properties of non-crystalline
materials: data and structure models" of the DKG (Wolfersdorf, September 1999);
b) Available on preprint-server cond-mat/0001319
130. Gastreich M, Marian CM, Gale JD (2000) Phys Rev B 62: 3117
131. Seifert HJ, Peng JQ, Lukas HL, Aldinger F (2001) J Alloy Compd 320: 251
132. Jalowiecki A, Bill J, Aldinger F, Mayer J (1996) Composites 27A: 717
133. Bill J, Kamphowe TW, Miiller A, Wichmann T, Zern A, Weinmann M, Schuhmacher J,
Miiller K, Peng J, Seifert HJ, Aldinger F (2001) Appl Organomet Chem
134. Bill J, Kienzle A, Sasaki M, Riedel R, Aldinger F (1995) Novel routes for the synthesis of
materials in the quaternary system Si-B-C-N and their characterization. Adv Sci
Technol 3B: 1291
135. Baldus HP, Passing G (1995) A new ceramic material for high performance
applications. Adv Sci Technol 7: 125
136. Wagner O (1991) Diploma Thesis, University of Bonn, Germany
137. Baldus HP, Passing G, Sporn D, Jansen M, Goring J (1997) Key Eng Mater 127/131: 177
138. DiCarlo JA, Dutta S (1995) Continuous ceramic fibers for ceramic matrix composites.
In: Lehmann R, El-Rahaiby S, Wachtmann J (eds) Handbook on Continuous Fiber
Reinforced Ceramic Matrix Composites. Ceramic Information Analysis Center, Purdue
University, IN, p 140
139. DiCarlo JA, Dutta S (1995) High temperature ceramic matrix composites I (HAT-
CMC-2). In: Evans AG, Naslain R (eds) Ceramic Transactions. American Ceramics
Society, Westerville, OH, Vol 57, p 141
140. Yun HM, DiCarlo JA (1999) Proceedings of the 23rd Annual Cocoa Beach Conference
and Exposition, Cocoa Beach, FLL, 25-29 January 1999
141. Kumagawa K, Yamaoka H, Shibuya M, Yamamura T (1997) Ceram Eng Sci Proc 18: 113
142. Takeda M, Imai Y, Ichikawa H, Kagawa Y, Iba H, Kakisawa H (1997) Ceram Eng Sci
Proc 18: 779
143. Ishikawa T, Kohtoku Y, Kumagawa K, Yamamura T, Nagasawa T (1998) Nature 391:
773
Author Index Volume 101
Aldinger F, see Seifert HJ (2002) 101:1-58
Friihauf S, see Roewer G (2002) 101:59-136
Herzog U, see Roewer G (2002) 101:59-136
Jansen M, Jaschke B, Jaschke T (2002) Amorphous Multinary Ceramics in the Si-B-N-C

System. 101:137-192
Jaschke B, see Jansen M (2002) 101:137-192
Jaschke T, see Jansen M (2002) 101:137-192
Muller E, see Roewer G (2002) 101:59-136
Roewer G, Herzog U, Trommer K, Miiller E, Fruhauf S (2002) Silicon Carbide - A Survey

of Synthetic Approaches, Properties and Applications. 101:59-136
Seifert HJ, Aldinger F (2002) Phase Equilibria in the Si-B-C-N System. 101:1-58
Trommer K, see Roewer G (2002) 101:59-136

Subject Index
ACC (amorphous covalent ceramics) 92, Calculation of phase diagrams (CALPHAD)

93,97-102 4,5,50,180
Acheson process 103,115 Carbonitride 3,5,28
Activation energy, of crystallization 99, -,/3-C3N4 3,5,28
100 Carbothermal reduction 103
Activity 21,34,40,49 Catalyst 69-71,72-75,79
Adamantane 88 Ceramic coating 169
Alkoxymethyldisilanes 71,72 Ceramic, digestion of 173
Amination 90,91 Ceramic, high temperature behavior 179
Aminolysis 153,158 Ceramic yield 61,62,67,68,73,77-79,82,
Ammonolysis 153,158 84-86,167
Amorphous ceramic 140,172 Chemical analysis 12,30,41,42
Amorphous covalent ceramics (ACC) 92, Chemical potential 13-15,21
95,97-102 Chemical vapor deposition (CVD) 61,
Amorphous silicon carbon alloy (a-SiC:H) 63-65,95,110,111,169
63,65,95,110 Chemical vapor synthesis (CVS) 113
CMC 169,170,183
Co-ammonolysis 154
Basic structure units (BSU) 95-97,101 Coating 169
Beta-hydride elimination Composite 153
(/J-hydride elimination) 73 Compound energy formalism 7
Boron carbide (C) 2,3,19,20,28,29, Concentration tetrahedron 3,5,45-47
41-50 Coordination spheres 175,176
Boron carbonitride 3,41,42 Creep behavior 185
-, BC2N 3,42,43 Creep resistance 183
-,BC3N 3,43 Crosslinking 62,67,68,77,78,86-89,91
-, BC4N 3,43 Crosslinking reagents 158,179
-,B2CN2 3,42 Crystal growth 95-102
Boron nitride (BN, B25N, B53N, B4N) 2,3,5, Crystallization 186
24,32-35,41-50 Crystallographic data 4,5
-, alpha boron nitride (a-BN, h-BN) 5,24, Curing 62,86,87,116,119,120,171
25,41-43 CVD see Chemical vapor deposition
-, beta boron nitride (/?-BN, c-BN) 5,24, CVS see Chemical vapor synthesis
25
-, gamma boron nitride (f-BN, w-BN) 5,
24 Dangling bond 90,94
-, high pressure phase diagram 25-28 Dehydrohalogenation reaction 143
-, phase transformation 25-28 1,3-Disilacyclobutane 77,79-84
-, turbostratic boron nitride (t-BN) 24 Disilane fraction (DSF) 69-72,116,117
Branching 73 Disilene 76
BSU see Basic structure units Disproportionation 69-72,91
196 Subject Index
DSF see Disilane fraction Inorganic network 141

DTA 99,100 Inorganic polymer 139
Invariant reactions 4,10,12,16,19,22,25
Isopleth see Temperature-concentration
Electrical conductivity 30 section
Electrochemical reduction 75,82 Isothermal section 29,31,32,36,37,45,
Electromotive force (EMF) 13 46,50
Electron probe micro analysis (EPMA) 4,
21
Electron spin resonance see ESR Kumada rearrangement 66,67,87,155
Enthalpy 13,14,16-18,21,23,25-27
Entropy 19,23
EPR (electron paramagnetic resonance) Liquid silicon infiltration (LSI) 185
88,90,93,97,100 Liquidous surface 29,33
ESR (electron spin resonance) 88,90,93, Lithography, soft 170
97,100 Long-range order 175
Ferroelectrics 139 MBE see Molecular beam epitaxy

Fiber, Si/B/N/C 182, 183 Medium-range order 175
-, Si/C/N 139 Melt-spinning 171
-, SiC 139 Membrane 124
Filter 124 Metallocene 74,75
Metallography 12,20,30
Metallorganic precursor 2
Gas species 4-6,28,38 Methane (CH4) 64,86,88,91
Gibbs energy of formation 17,18 Methylchlorodisilanes 69-71
Gibbs free energy 5,6,17 Microstructuring 170
GPC 82,83,164 Molecular beam epitaxy (MBE) 109
Gravimetry 12 Molecular mass 72-80,82,84-87
Grignard reaction 144 Molecular mass distribution 67,75,76,164
Molecular weight see Molecular mass
Muggianu algorithm 7
Heat capacity 12,13,16,18-21,25,26 Multinary systems 138
High resolution transmission electron
microscopy (HRTEM) 47
High temperature stability 47 Nanocrystalline phase 180
Homogeneity, spatial 176 Nicalon 139,184
Homogeneity range 3,5 NMR (nuclear magnetic resonance)
- -, boron carbide 3,5,19,20,22,29 82-84,87-92,97,98,101,102
- -, silicon boride (SiBn, SiB6, SiB3) 3,5,7 -.solid state 164,167,174
Hydroboration reaction 143,155,156 Nucleation 95-102,109,113,114
Hydrogen elimination 154,156
Hydrosilylation 72,73,84,85
Oligosilane 69-71,77,91
Ostwald ripening 100
ICP-OES 173 Oxidation behavior 181
Infiltration 115
-, chemical vapor infiltration (CVI) 116,
123,125 Pair correlation function 178
-, liquid polymer infiltration (LPI) PBS-Me 163
116-118,123,125 PD see Polydispersity
-, liquid silicon infiltration (LSI) 117,123, Phase diagram 4
125 - -, B-C 19-23
Infrared spectroscopy 167 - -, B-C-N 41-44
Subject Index 197
- -,B-N 24-28 , Si-B-C-N 45,48

- -,Si-B 7-10 , Si-B-C 31,34
- -, Si-B-C 28-34 , Si-C-N 38,39,40
- -,Si-B-C-N 45-50 Scientific Group Thermodata Europe
- -,Si-B-N 32-36 (SGTE) 6
- -, Si-C 10-15 Segregation 180
- -,Si-C-N 35-41 Self-diffusion 139,152
- -,Si-N 15-19 Semiconductor 63,65,107
Phase equilibrium 2-5,7 Short-range order 174
Phase fraction diagram 36,39,49,50 Si2CN4 3,5,6,36
PMPS see Polymethylphenylsilane Si3N4 139
Polycondensation 153,166 SiB3N7 3,34
Polydispersity (PD) 72-79,82 SiC (silicon carbide) 139
Polymer film 63 Sigma-bond metathesis (F-bond meta-
Polymer infiltration 169 thesis) 74
Polymer route 139 Silacyclobutane 77,79-82
Polymethylphenylsilane (PMPS) 68,76 Silaethene 67
Polysilaethene (PSE) 77,82,83 Silane 64
Polysilastyrene 67,68 Silazane cleavage 141
Polysilyne 68,69 Silicate glass 93,94
Potential phase diagram 15,19,34,36, Silicide 102
40-43,47-50 Silicon boride 3,5,7,32
Powder route 152 - -, SiB3 3,5,7
Precursor-derived ceramics 4,34,36,37, - -, SiB4 7,29
39,43,45-47 - -,SiB 6 3,5,7
Protective double layer 181 - -,SiB n 3,5,7
PSE see Polysilaethene Silicon carbide (SiC) 2,3,5,10,13,14,28,
Pyrolysis 61-66,73,76,77,85-102, 29,35-41,44-50,124,125
111-113,116 -, alpha silicon carbide (a-SiC) 5,6,10,
-, atmosphere 166 13-15
-, monitoring 165 -, beta silicon carbide (/?-SiC) 5,6,10,
-, temperature program 166 13-15
-, carbon composite 124
-, crystallization 90,95,99,101,102
Quantum mechanic calculations 176 -, doping 63,65
-, fiber 61,106, 111, 112,115,117-120
-.film 65,111
Random network 172-176,186 -, foam ceramic 126
Reaction paths 37 -, high temperature application 114-127
REAPDOR 175 -.layer 108-111
Redlich-Kister polynomial 7,22 -, nano-scaled powder 103-106,112-114
RED OR 175 -, phase transformation 10
Reflection high-energy electron diffraction -.platelet 115
(RHEED) 109,110 -, polytype 102,108,109
Rheological properties 164 -, porous 124,125
Rice husk 104 -, precursor 63-85
Ring opening polymerization (ROP) 75, -.properties 114
79,83 -, reinforced glass 120-123
-, single crystals 106-108
-, superlattice 109
Salt elimination 143 -, surface 108
SAXS 98,99 -, surface reconstruction 110
Scheil reaction scheme -, whisker 104,111,112,115
,B-C-N 43,44 Silicon carbonitride, beta-(SiC2N4) 3,5,6,
,Si-B-N 36 36
iy» Subject Index
Silicon nitride (Si3N4) 2-6,15,32-41, - -,B-N 24,25,28

44-50 - -, Si-B 9,10,11
-, alpha silicon nitride («-Si3N4) 5,6,15 - -, Si-B-C 29,31-34
-, beta silicon nitride (/?-Si3N4) 5,6,15 - -, Si-B-C-N 45-50
-, cubic silicon nitride (c-Si3N4) 5,15 - -, Si-B-N 35
-, phase transformation 15-19 - -, Si-C 14,15
Silyl triflates 78 - -, Si-C-N 36,37
Silylene 70 - -,Si-N 16
Single source precursors 146-151,154 Thermodynamic experimental data, B-C
- - -, molecular structures 152 21,22
- - -.syntheses 141-145 , B-N 24-28
Soft lithography 170 , Si-B 9,10
Sol-gel process 154 , Si-B-N 34
Solid solution phases 2,5 , Si-C 12-14
, boron (/3-B) 5,7 , Si-N 15-19
, boron carbide (B4+6C) 5,7,19,20, Thermodynamic modeling 5
22,29 Thermodynamic optimization 6
, silicon boride(SiBn,SiB6,SiB3, Thermolysis 2,3,35,50
SiB2.89) 5,7 Transamination 153
Solid state reaction 139 Transition metals 72,73,78
Stannazane cleavage 142 Transition reaction 31
Stoichiometric composition 63,77 Transmetallation reaction 143
Structural ceramics 138
Sublattice model 10,22,29
Vapor pressure 18
Vapor-liquid-solid mechanism (VLS)
Temperature-concentration section 112
(see Isopleth) 29,32,33,36,44,50 Vapor-solid mechanism (VS) 112
TG/MS 88,89,91 Viscosity 164
Theoretical studies see Quantum mechanic VLS see Vapor-liquid-solid mechanism
calculations VS see Vapor-solid mechanism
Thermal analysis 4,12,20,30,40,50
- -, differential scanning calorimetry
(DSC) 4 Wood 105,106,117
- -, differential thermal analysis (DTA) 4 Wurtz coupling 66-69,76,77
- -, thermogravimetry (TG) 4,40,46 Wurtz reaction type 141
Thermodynamic calculation 4,5-7
- -, B-C 22,23
- -, B-C-N 43 X-ray diffraction 4,12,20, 30,36,40-43
Printing (Computer to Film): Saladruck, Berlin

Binding: Stiirtz AG, Wiirzburg

High Performance Non-Oxide Ceramics I

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

High Performance Non-Oxide Ceramics I

Caricato da

Copyright:

Formati disponibili

101

Structure and Bonding

Volume Editor: M. Jansen

As a rule, contributions are specially commissioned. The editors and publishers

Springer WWW home page: http://www.springer.de

CIP Data applied for

Springer-Verlag Berlin Heidelberg New York a member of BertelsmannSpringer

Typesetting: Scientific Publishing Services (P) Ltd, Madras

Professor D. Michael P. Mingos Professor Dr. Martin Jansen

Prof. Allen J. Bard Prof. Ian G. Dance

Prof. Peter Day, FRS Prof. James A. Ibers

We are pleased to announce that Professor Michael Mingos, University of

Dr. Marion Hertel

Phase Equilibria in the Si-B-C-N System

Silicon Carbide - A Survey of Synthetic Approaches, Properties

Amorphous Multinary Ceramics in the Si-B-N-C System

Author Index Volume 101 193

Subject Index 195

High Performance Non-Oxide Ceramics

Boron Nitrides - Properties, Synthesis and Applications

Silicon Nitride Ceramics

Literature information on phase equilibria and materials thermodynamics of the quaternary

Keywords: Phase equilibria, Thermodynamics, Si-B-C-N system, Precursor-derived ceramics,

2 Stable Solid Phases in the Si-B-C-N System 5

Structure and Bonding, Vol. 101

6 The Quaternary Si-B-C-N System 45

necessarily requires a knowledge of the multicomponent phase equilibria

(including numerous gaseous species). Although the metastable phases

Stable Solid Phases in the Si-B-C-N System

The compositions of solid phases appearing in the Si-B-C-N system are

The CALPHAD method of computer coupling of phase diagrams and

Silicon cF8 Fdim C(dia.) 0.54309

The descriptions of all ternary systems were combined in one dataset to

Table 3. Experimental phase diagram data of the system Si-B

Metallography Microstructure T < 2373 0-1.0 Knarr (1959) [43]

Diffusion Solvus 1073-1286 0.998-0.999 Schwettmann

accepted to be formed at about 1543 K by a peritectoid reaction [60]. Aselage

Table 4. Experimental thermodynamic data of the system Si-B

1600 - SiBg + (Si)

Table 5. Invariant reactions in the system Si-B

Reaction Composition Composition Composition Temperature

1+(B) = SiBn (p) 7.4 2.1 3.3 2310

Table 6. Experimental phase diagram data of the system Si-C

Type of experiment, Type of Temp. Cone. Literature

Gravimetry Liquidus 1793-1998 0-0.01 Hall (1958) [78]

Table 7. Experimental thermodynamic data of the system Si-C

Type of Temperature Phase Literature

enthalpies of formation [90, 98-100] and chemical potential measurements

Liquid Liquid + Graphite

Olesinski (1984) [88] and Kleykamp (1993) [85], respectively. Vaporization

Experimental thermodynamic investigations of the Si-N system mainly

Fig. 4. Chemical potential of silicon in the two phase region SiC-C

Table 8. Invariant reactions in the system Si-C

Reaction Composition Composition Composition Temperature

G = l+C 58.4 78.7 0 3216

An assessment of the Si-N system was presented by Carlson (1990) [114].

Table 9. Experimental phase diagram data of the system Si-N

Type of Type of data Temp. Cone. Literature

Table 10. Experimental thermodynamic data of the system Si-N

Type of Temperature Phase Literature

Si3N4(s) = 2N2(g) (1)

According to the study of the vaporization behavior of Si3N4 [125] besides

Fig. 6. The Si-N potential diagram (log p N2 -T) [272]